The Northern Spy
June 2003
Roundup and Rumour
by
Rick Sutcliffe
|
|
In May, the Spy commented at some length on NisusWriter and Word. However, there are other word processors available for the Mac OS X platform.
Mellel, from Tel Aviv based RedleX (www.redlers.com) is a relatively recent (Jan 2002) low-cost (currently $19, to be $25; specials on quantities) word processor that is currently in active development with numerous essential features already implemented and more on the way. It's a refreshing throwback to see a company that is not afraid to innovate, to move the envelope from what is viewed as canonical for its product category. To cite a couple of instances, Mellel ignores Apple's lack of support for right-to-left text display, and just does it anyway, thus the developer is able to claim this product is the only one offering Hebrew word processing on OS X. NisusWriter offers it in the original classic version, but not yet in Express.
An example of innovation: dynamic spell checking can be done on a per paragraph basis, rather than as each word is typed (user option). This means that one can set the spell checker to wait until return is pressed, then check the entire paragraph. Another: The user can set a secondary font with characters relative to the primary font. If the primary font size, say, is reduced, the secondary one's size would also diminish accordingly. A third: changes to styles are live, that is, one need not wait until after closing a window to see the results. Yet another: The separation of page, paragraph and character styles make for much more versatile document construction.
Mellel features include:
* Page, paragraph and character styles
* Tables
* Footnotes and endnotes as part of an enhanced "note tool"
* Advanced spell checking options
* Headers and footers
* Live attribute changes
* RTF and text import and export
* Innovative tabs for indenting nested lists
* Multilingual support including Greek, Spanish, Hebrew, Cryllic and French
* Image placing and resizing
* Easy insertion of special characters, control characters and variables
* Right-to-left writing support
* Real-time word count
* Automatic save
To come (per the website): outline tool, index, table of contents, columns, long-document management tools, Arabic support, and more.
What you can't do: read or write Word (except via RTF) or export HTML or PDF. There is no Classic version, invisible text format, mail merge, thesaurus, GREP in find, and there are no explicit rulers or macros.
However, Mellel has the look and feel of a finished product rather than the rough edges of a work in progress. The user interface is well thought out, the initial feature set sound, the plans for the future promising. When I corresponded with them about invisible text, they were available, interested, and engaging. In the end they added this to the to-do list.
With this feature, the ability to read and write Word documents, macros, and GREP in the find command, Mellel would be an excellent product. It's already very good. An impressive start.
Mariner Write from Mariner software has, by contrast, been around for some time. In fact the Spy has tried several times before, the last a number of years ago. From Mariner software, the program is $70 by download or $10 more for a CD.
One of the main claims to fame of Mariner Write is a small memory footprint (2 M of RAM, or a twentieth that of Word), and it is reputed to be fast, though any program could make that claim beside the glacial Word.
However, Mariner Write didn't make as good an impression out of the gate, because it unpacked into an application folder that when clicked on appeared to be incorrectly constructed, as then clicking the application just brought up another finder window. This problem went away on reboot, but it has happened before with new applications, and is apparently a byproduct of the way they are packaged.
Because of its long heritage, Mariner Write is available in both Classic and OS X versions. It also has more of the basic features, such as a thesaurus, mail merge and the ability to read Word documents. It can also save in RTF or PDF, has an invisible (hidden) text format, mail merge, tables, paragraph and character styles, and a thesaurus. It too has dynamic spell checking, though with fewer options than Mellel.
Mariner Write features Include:
* Automatic spell checker highlights mistakes as you type.
* Contextual editing and spelling suggestions.
* WorldScript savvyÐsupports Kanji and numerous other international scripts, provided Apple does.
* AutoSave feature lets you work without worrying about losing your file.
* WYSIWYG font menus let you see the typeface before using it.
* Save your documents in PDF, RTF, and various ASCII text formats.
* Automatic save
What's missing? The ability to write Word or HTML, GREP in the find command, right to left text, and Macros.
That is, as things stand, Mariner Write is a more complete product than Mellel. It is also a more conventional one, a kind of middle-cost alternative to Word, with a minimum of innovation or new ideas. Given the length of time it has been around, this may not bode well for a competitive future, though the company has done a good job on the X version, especially considering the product was all but moribund for several years .
Invisible text and Markup: As mentioned last month the Spy uses it in NisusWriter to store html markup so that documents will both print correctly and display the same or a similar formatting when viewed as web pages. It is also useful to temporarily remove material from view in a document or to comment on the document. (Word's revision feature is better for the latter, however, one of its few good points.)
When he creates small files for the web, he uses BBEdit and its macros to insert code. But for his shareware textbooks, so as to have just the one version both print and display, he uses NisusWriter, and in Classic has also employed publish-and-subscribe heavily.
Don't apply the Nisus styles and then tell it to generate HTML because the result is poor, almost as bad as the html that Word exports. Indeed, our experience with html generators is none too good. Neither is that with wysiwyg html code generators, which tend to create unnecessarily bloated results. (As a web host,--another hat--the Spy allows Front Page produced sites, but don't support them, because FP code is very complex and often works with only a single browser.
The Spy writes with two or three basic paragraph styles per book plus a lot of markup bold, italics, font size changes, and so on. This requires the best of both worlds, text formatting as only a word processor can do, and web display. Nisus gives the best of one world, and a half-baked solution (almost working invisible text) for the other. BBEdit (and the web page creation software) gives only the second.
So, why not have a word processor that automatically exports HTML. Then one could turn any page into a web file. After all Word does this, so how hard could it be? Well, Word doesn't do it very well, and frankly, the Spy is not optimistic that an automatic generator could produce terrific html, nor that a product will ever do both well.
Take for instance the display of code (i.e. programs written in Modula-2 or Python). One tends to have several layers of indenting, colour coding and/or bold to mark reserved words, etc. The indenting problem virtually forces one to use blockquote tags to get a correct display, though within that there would still be character styles to mark reserved words, say.
The Spy guesses that one could write a decent html export utility for Mellel because of the clear separation between character and paragraph styles. For the former, simple bracketing markup tags would do. For the latter, placing a p.style in an associated CSS could work for many things.
Tables would be easy, and the meta tags for a file could be user-configured in a separate file with the same name and a different extension, and imported when the final HTML is generated. However, this would only give the most basic document, not exploit the full formatting power of HTML (or of the word processor for that matter.) There's a market opportunity here. The first manufacturer that could produce a product that is wysiwyg both as a word processor and as a web site generator could make a killing. Well, the Spy would buy a copy anyway.
The Last Word Neither Mellel nor Mariner Write is any more ready to replace NisusWriter Classic on my productivity desktop than is Nisus Express for X. However, given time and resources, all three have the potential to be better products than the slow and bloated Word, especially for book length projects, which Word is utterly incapable of handling. The Spy needs a professional writing tool that works well with up to thousand-page documents.
But for scribblers of small documents, all are already viable alternatives. For the long run, I suspect Mellel may have the horses to become a truly great product, especially if the pace of innovation continues. It is a "write different" product. OTOH, NisusWriter has the pedigree to become great again, perhaps even to surpass its venerable but aging parent, but it's taking a long time. Will the market wait?
To ride another hobby horse, will the code be written the way we tell our students to at TWU--well-documented and modularized, with system dependencies firewalled in their own compartments? Or will they be clever hacks in write-only code, flashes-in-the pan that last as long as it takes the would be creators of the next version that they don't understand the old code, even when it's their own? (This is a good description of many once-great products that never got the next version out the door, and an apt one for at least parts of one of those discussed here.)
Perhaps the most instructive aspect of this is that at least there are alternatives, that there is development activity. Both tell us that a new Macintosh word processor is not only desirable but also possible. That both sell multi-language capability hints at why new products are needed. There is a wildcard, however.
What is Apple doing? Given that Apple already has productivity tools in several market segments, and is opening new ones at a breathtaking pace, one must speculate on whether they will eventually offer their own office suite. Only a spreadsheet and word processor would be necessary to go into direct competition with Microsoft on this front. WWSD? (What Will Steve Do?) Apple has the resources to develop both products in house, and the jam to bring both to market without any advance leaks. They already have a top flight database program. Now, a year ago, one might have said that a made-to-order word processor was about as sensible as Apple offering their own browser. Indeed.
NisusWriter, Mariner Write, and Mellel all demonstrate that it can be done. Mellel thinks most differently. Apple could buy and complete any one of these three, or roll one of their own.
Perhaps the real fly in Apple's ointment if going this route would be a spreadsheet product. There have been several attempts to go head to head with Excel, including Mariner Software's own Mariner Calc. However, the Spy has never found any of the alternative spreadsheet programs to be of much use. His own spreadsheet work (direct descendents of sheets first created in Visicalc) is so heavily dependent on macros that products lacking them are of no value to him. So far, this one is game, set, and match to Excel, Microsoft's only (IMHO of course) best-of-category software (neatly matching Access which holds down the other end of the scale in its category.)
Note: The Spy's comments on word processors should be taken in the context of his experience with dozens of word processing and text creation tools, on computers of all sizes and shapes going back to the sixties (and a few dedicateds). He does not know how many he has used, but guesses the number is more than fifty. Moreover, and here's the disclaimer, this is a general technology review, not a detailed feature comparison. The Spy does not claim to have exhausted the interesting points on any word processing product mentioned here, just to have chatted about what strikes his fancy, or doesn't.
Other grist in the rumour mill Persistent stories aver that Apple has already taken delivery of production-run quantities of IBMs 970 chip in 1.4GHz through 1.8 GHz speeds. If true, Apple could have more than just Panther to reveal at WWDC, because there appears to be no reason why production units could not be available to show by late June. Such an revelation would be a departure from normal practice, as Apple prefers more public venues to make such announcements.. However, Steve owes developers something after bumping WWDC over by a month and a few cities. The demo of a new desktop would go a long way to paying off that debt. Giving a couple away at random to developers would be another.
What's so special about the 970? It's the next generation, a 64-bit RISC chip that is a desktop cut-down from IBM's Power4 server chip, but done in such a way as to be ideal for Apple's needs. It will run current 32-bit code, with plenty of room for speed increases as chip speed performance and 64-bit code optimization ramp up.
The chip also makes sense for a laptop because, at 1.2GHz it consumes a mere 19W. The 970 has a transistor count approaching that of a P4, while still using RISC architecture. It features twelve execution units (including two FPUs), faster bus speed, more instructions fetched per cycle (eight) more sophisticated pipelining, a larger instruction cache (twice the G4's), more real estate devoted to branch prediction up to 200 instructions on board simultaneously--all permitting higher throughput for a given clock speed.
The big challenge for Apple is to get a high-bandwidth frontside bus and memory architecture into its mother boards, whose engineering has not even been up to the capabilities of the chips they've been using. Continuing with inadequate motherboards would be an embarrassment at this stage of things. (OK it already is, but...)
The chip is designed to be multiprocessed, and an XServe box with, say, four or eight CPUs would not be difficult to architecture from a basic unit, making Apple a much more credible player in the higher-end server market. For various technical reasons, the 970 will likely be a relatively better floating point and vector processor than an integer processor, meaning other high-end applications like gaming will benefit disproportionately. (Note that what Motorola calls AltiVec is jointly owned by them with Apple and IBM and has been added to the 970, a sure sign that the chip was designed to Apple's specs as a G4 replacement and G5 killer.)
What this means is that the Apple-Motorola relationship is all but finished and a new era of Apple-IBM cooperation has begun. Assuming a satisfactory motherboard design, (Apple's engineers have to have been doing something these last few years!) we ought to see a completely new desktop line topping out at dual 1.8GHz towers initially, portables in the 1.2-1.4GHz range a little later in the year, and new dual/quad/octal servers by the fall. There are too many variables to accurately predict comparative throughputs, but given a new bus, 970 chip, and faster memory, these machines ought to at least double the performance of existing code, and perhaps nearly triple with properly optimized code. (64-bit architecture does not in itself necessarily produce faster throughput; chip and motherboard architecture accomplish that.)
The Spy cannot help but wonder though. Can Apple pull off a double coup? Will Panther be available in a 64-bit version? WWDC attendees would forgive a lot for an announcement like that one.
-- The Northern Spy
Websites referred to (some bumph copied from and edited):
www.redlers.com
http://www.marinersoftware.com/
http://www.theNorthernSpy.com
http://www.arjaybooks.com
|