No, this is not some Zen koan: for a long time to come there will be two ArcViews, the new and the old.The new ArcView will have a different interface and different name but will provide the same functionality and has the same look to its output.Underneath, though, it is a completely new architecture.

"The majority of ArcView users," says one ESRI representative, "know their twenty clicks:" how to load their data, how to put a symbol in a legend, how to print, and so on."It's a tool.For these users, once they get over the differences in GUI [graphical user interface], the new technology will be as good or better."

The new ArcView has been under development for five years, almost since ArcView 2.0 emerged as a powerful analytical tool for geographic data in 1995.Back then ESRI architects foresaw the need to improve and isolate key parts of the code within independent "engines." There are now a grid engine, network engine, geocoding engine, print engine, and others.

These engines form the underlying code to ESRI's new "Arc" products: ArcMap, ArcCatalog, ArcToolbox.This fall, ArcView users will be given the choice to upgrade to this software suite.Word is that the upgrade will cost around $600.

(ArcToolbox capabilities will be limited, to differentiate the new ArcView from the more expensive suite to be sold as ArcInfo 8.1.ArcMap and ArcCatalog, however, will be the same products sold with ArcInfo.Eventually, ESRI will also port many of its current ArcView extensions.Spatial Analyst has already been ported and the rest have been "prioritized" for porting.)

ESRI will still sell and support ArcView as it presently exists.Now in version 3.2a, it may soon see an inexpensive upgrade to version 3.3.Todd Stellhorn, lead developer for ArcView 3.x and its supplemental "extensions," claims this is a substantial upgrade.He cites support for AutoCAD 2000 files, improvements in some data and image readers, bug fixes, and a "couple of new Avenue requests." (Avenue is ArcView's scripting language.A new Avenue "request" can be thought of as an addition to a body of related subroutines built into this language.)

Many within ESRI are quick to point to the aging PC-ArcInfo product as an example of their software's longevity.That product is now several generations old but is still being used by many customers.ESRI has long ceased to improve PC-ArcInfo but they still sell it and support its users.

It appears that ArcView 3.x is headed in the same direction.ESRI recently released two new extensions--Image Analyst 1.1 and Model Builder--but has "no firm plans to further develop new extensions," according to Stellhorn.

ESRI representatives acknowledge that many users are apprehensive about the changes.Off the record, they suggest user perception is related to change management.Some customers, they say, have been through major upgrades before; they have the resources to learn the new product and migrate their data and users to it without interrupting their work.Other customers simply "have no idea" what to do.

Third-party developers are in a similar position.They have had to learn the ArcView scripting language, "Avenue," which presently supports over 900 types of objects and over 2000 "requests"--executable procedures--for those objects.They can use this experience to develop products for an aging and shrinking base of ArcView 3.x users, or they can learn to program in the new system.

There are many ways to program in the new system, but none of them use Avenue.Programming will be COM-based, implying developers will have unprecedented capabilities and a choice of programming languages (Visual Basic and C++ will be the most popular).The scripting language will be VBA (Visual Basic for Applications).

Avenue's 900 objects have morphed into over 1200 classes in the new ArcView.ESRI programmers proudly point out this is "the largest COM-based application in the world." Thad Tilton of ESRI asserts, "If you can program this, you can program anything [in COM]." That may be small comfort to many developers standing at the bottom of this learning curve.

One ESRI programmer advises, "just get a book on programming VBA in Microsoft Office.Then you'll be all set." He acknowledges there are still the 1200 classes to learn, but at least it's a start while waiting for the new software to be released.

For those users or developers trying to decide between the old and the new, there is another consideration.Despite its maturity, ArcView code still contains niggling bugs.The obvious ones are in the "geoprocessing" code.This is the code that compares and modifies map features by intersecting them, buffering them, and merging them.ESRI is working on fixing this, but for about two years to come, says a source within ESRI, "there will still be some ragged edges."

In the balance, though, the new ArcView has to be preferred.Its potential for further development, improvement, and customization is awesome.The improvements it has, even in its present "beta" test release, are substantial.The question for present users is not whether, but when, to upgrade.The old ArcView is dead.Long live the new ArcView.

--William Huber
San Diego, June 29, 2000.