(Part Two of this article is available here.)
Rich Turner is the Product Manager for desktop software at ESRI in Redlands, California. We spoke with him to follow up on our reports of the ESRI 2000 User Conference held in June.
During his career at ESRI Turner has helped launch PC ArcInfo and the ArcView “desktop GIS” products. With the imminent convergence of the underlying technologies in ArcView and ArcInfo, Turner sees his role “morphing into ArcGIS” product management.
We asked Turner for some forward-looking, strategic thoughts about the future of ArcView. A natural place to begin is with one of the most obvious changes afoot: the move to a Windows-based architecture. ArcView has been a multi-platform product from its inception. The new ArcView, version 8.1, is built on Microsoft’s COM technology, which runs only on Windows.
There are benefits to Windows compliance on the client side. Turner rattles off several in quick succession: drag and drop, Windows standard applications, interoperability with other Windows apps, and the availability of well-known developer tools for customization. In short, better integration with existing software investments and greater access to programming talent and knowledge.
Apparently, however, a substantial number of ESRI’s customers run ArcView on non-Windows platforms. When asked about how these users will be supported, Turner gave a three-part answer. You might characterize these as “follow the leader,” “port and pray,” and “golden oldie.”
The first answer carries echoes of a grand change in software architecture that is slowly unfolding. No longer does one machine do all the computing, as mainframes often did through the 70’s and PCs did in the 80’s and early 90’s. Client-server and distributed computing technologies allow a “server” computer to manage large databases and perform intensive computing operations. A “client” device (which for many people soon will be hand-held, mobile, and wireless) provides a user interface and small amounts of local computing.
Client machines, which include PCs and laptops, usually run a version of Windows. On the other hand, Windows NT and varieties of Unix compete for dominance in the server category. The small mobile and wireless machines--“thin clients”--often run a platform-independent application written in, say, Java.
ESRI is writing its GIS software to match these configurations. “The basic philosophy is to match our product strategy to the appropriate platform, as defined by the majority of our users, which is the client on Windows and the server on Unix and Windows NT,” says Turner. The “rich clients,” such as the forthcoming ArcView 8.1, will be written to run only under Windows. This is the follow-the-leader strategy.
Supplementing this strategy is an effort to port the supporting COM technology from Windows to another operating system, Sun Unix in this case. “There’s good news and bad news” about this, says Turner. The good news is that ESRI is doing the port and that “something will happen.” The bad news concerns the unavoidable constraints.
“COM” is Microsoft’s Component Object Model. The appellation says it all: anything with Microsoft in the name is going to run under Windows, not Unix. ESRI can port its components (some 1200 of them at last count), but it cannot port the third party COM objects it licenses to round out its new GIS software. These third party objects include report writing, charting, and even VBA (Visual Basic for Applications). Without a port of VBA, it is difficult to create a customization environment for ArcView. ESRI is hoping Microsoft and others will go along and port their objects, too. This is the “port and pray” aspect of ESRI’s strategy.
Incidentally, this effort is not a software emulator, but a true port: a rewrite of the software to run under a different operating system. Turner frequently uses the word “research” to describe this effort. Read that as meaning not to expect too much too soon, but keep an eye out for news.
The third part of ESRI’s strategy is well known. They will continue to support ArcView 3.x on Unix. According to Turner, “the market will decide” the future of this path. Like PC ArcInfo before it, ArcView will persist as a viable but aging technology, a “golden oldie.”
So, for the near future at least, the price of admission to the new ArcView will be to run Windows (98, NT, or 2000) on your computer. Who will be the early adopters? ESRI, of course, would like everybody to upgrade to the new technology. “Our goal for the initial release is to achieve feature-equivalence or better with ArcView 3.x, so that ArcView 8.1 is an even more compelling product: you get more of the same features, only some are even better.”
Some of the improvements Turner cites include on-the-fly projection (which, we understand, includes projection of images and raster data sets), rotation of maps in layouts, VBA in place of the proprietary Avenue scripting language, and support of annotation in the database. These are features many users have been asking for. You just cannot do these things in ArcView 3.x.
However, there are reasons why some current users might want to wait to upgrade. These are the people who rely on “extensions” to ArcView--that is, supplementary software--to perform critical tasks. Important extensions include Spatial Analyst, which provides raster analysis, 3D Analyst, Image Analyst, and Network Analyst.
Five extensions are slated for concurrent release with ArcView 8.1. In addition to Spatial Analyst and 3D Analyst, expect to see StreetMap, ArcPress, and the new Geostatistical Analyst (a sophisticated tool that’s “not for Bubba, but maybe for Bubba’s consultant” joked one developer at last month’s conference).
Turner asserts that ESRI is rewriting these extensions, not just repackaging old code, so expect a new mix of capabilities and some enhancements. What’s interesting is that capabilities formerly available only in high-end (that is, pricier) extensions like Image Analyst and Network Analyst are “now in core, partially.” This means a broader range of capabilities and tools will be available to all ArcView users. Some customers might not have to purchase or upgrade extensions if their needs are basic.
Other customers may want to use both versions of ArcView. “We don’t want people to think it’s an either-or scenario. Even for people who depend on extensions [who are therefore those most likely to stick with ArcView 3.x], there is still a benefit from using 8.1,” says Turner. Some benefits of this hybrid approach: ArcCatalog for data management and metadata maintenance, new “high-end” cartographic capabilities, and new editing tools.
Users considering whether and when to upgrade to ArcView 8.1 will have to take the changes to the core software into account. First, take a close look at the base product: it may already have all the features you need, even if you currently use one or more extensions. If it does not, possibly one of the first five extensions will provide those features. In those cases, upgrading soon makes sense. Otherwise, watch for new capabilities (especially with image processing and “temporal data management”: change detection, mapping polygons and raster images over time, and so on) to be announced by the User Conference in mid-2001. Until then, adopt the “golden oldie” strategy and stick with version 3.x.
(Part Two of this article is available here.)