From Online Reporter, a sister publication Can Microsoft transform the medium of the World Wide Web from one that is essentially platform independent to one that can only be viewed effectively from a Windows machine? With ActiveX it is having a good go, according to our sister publication Online Reporter. Few people seem to have […]
From Online Reporter, a sister publication
Can Microsoft transform the medium of the World Wide Web from one that is essentially platform independent to one that can only be viewed effectively from a Windows machine? With ActiveX it is having a good go, according to our sister publication Online Reporter. Few people seem to have noticed, or at least commented on, Microsoft’s sleight-of-hand in introducing ActiveX controls as a Web technology. The fact that ActiveX controls will need to be recompiled to run on non-Wintel platforms has hardly been mentioned. The fact that Web site owners will need to keep multiple copies of each ActiveX component – one for each browser platform – hasn’t been murmured. Even when the ActiveX control is written totally in Java, with just a skinny ActiveX covering, it is likely to need recompiling; so the benefits of coding for Java in the first place are lost.
Mike Pryke-Smith, Microsoft’s Internet tools product manager, says that it is too early to know this for sure but my hunch is it will need a recompile. An ActiveX object is a binary COM object, no matter how much Java is packed into its heart. It will need to be recompiled to match the COM implemention in the non-Windows browser. Microsoft’s big claim to cross-platform compatibility has been based on its intention of getting Internet Explorer 3.0 out for Unix and the Macintosh by the end of the year. All the publicity and material from the company imply – and sometimes state – that these versions will run ActiveX controls. This is true. The problem is that they won’t run the controls that are shipping today. Those will need to be recompiled first , and that’s not inexpensive or trivial. Microsoft is working with Metrowerks to get ActiveX working on the Mac. As for Unix, it is working with two companies: Bristol Technology in Ridgefield Connecticut and San Francisco-based MainSoft. Both are Windows source code licensees, both are in the business of providing tools that let developers recompile their Windows source code to run on various flavors of Unix. As such they offer fully featured cross-platform development tools for moving between Wind ows and Unix – and fully featured doesn’t come cheap. Bristol’s Wind/U package, for example, costs anywhere between $5,000-$9,950, depending on volume. Each copy of the package produces binaries specific to a particular flavor of Unix. Let’s assume that the ActiveX developer would like his work to be seen by browsers running on Sun, HP, DEC Alpha, AIX, and SGI machines – are they going to want to pay $25,000? This isn’t to criticize Bristol; the tool was designed for large-scale application porting, not the corner-shop developer who wants to make an animation applet work on all browsers. Bristol’s director of strategic relations Chane Cullens says the company is considering whether to produce a stripped-down tool just for this purpose. But since there won’t be any demand for the tool until the Unix version of Internet Explorer ships, a firm decision has yet to be made. Assuming that ActiveX controls do get recompiled en masse, and assuming that Explorer becomes available across multiple operating systems, there is still the browser question. Netscape is currently adamant that it won’t support ActiveX – its depth of feeling on the subject was demonstrated recently with the announcement of the Netscape ONE platform which specifically excluded it. Microsoft has circumvented this dislike by co- developing a plug-in with NCompass Labs in Vancouver. NCompass ActiveX Plug-in Pro bridges the competing worlds of Netscape’s LiveConnect and ActiveX. But NCompass says it has no plans to port the plug-in so that it works with non-Windows Navigators. Still it is just possible that ActiveX controls will one day get Java’s write-once-run-anywhere ability. Microsoft’s Pryke-Smith says there’s the possibility of using a virtual machine-type approach in future versions of Explorer. In the same way that Java code is interpreted today, ActiveX code would be interpreted within the browser platform. Pryke-Smith admits however that this is a long way off. The bottom line then. De velopers who use ActiveX components in their Web pages are currently restricted to having those pages viewed by Windows clients running Explorer or Navigator plus plug-in. Even when Explorer becomes available on other platforms they will have to explicitly recompile and distribute variants of the control for each browser platform. The upside is that in a Windows-dominated world, ActiveX offers faster and more flexible access to the browser’s resources. Companies that have already sunk investme nt in OLE can re-use their code on the Web with a minimum of effort. Microsoft likes to quote figures from the Giga Information Group showing that $240 million of components will be sold in 1996, to grow to $1 billion by 2000. But without inexpensive recompilation tools and cross- browser support, ActiveX could transplant the Windows hegemony from the desktop to the Web. That might not be a bad thing from Microsoft’s point of view, but it will be well worth watching how the prospective independent ActiveX standards group will handle the issue. And does it really fit with the ethos of the Web?
By Chris Rose