IE – app or OS? Microsoft case grinds to another weekend

The latest might be a narrow victory for the DoJ, but it was largely adrenaline-free

Fruitless skirmishes between Professor David Farber of the University of Pennsylvania and Steven Holley for Microsoft continued for a second day yesterday, but gained no points for Microsoft. Farber said that Hadi Partovi, a Microsoft employee, had said in testimony that IE was an application. It turned out that Partovi had said that explorer.exe was an application, but Holley claimed that the executable "does nothing but call other files in Windows 98", but it still looked as though Partovi had made a Freudian slip. There was extensive questioning as to whether Microsoft had put several functions into one DLL in order to make it impossible to exclude a DLL from Windows 98 by ensuring it had a Windows 98 function as well as an IE function. Farber did not know enough to prove the point, but the suspicion lingered. Holley almost admitted that some large program elements might have multiple functionality when he criticised Netscape's "monolithic" navigator.exe, seeming to use again the argument that everybody does it, so therefore it's excusable for Microsoft to do it. This was a particularly weak remark by Holley, because there was no suggestion that Netscape had combined functionality for the same reasons that Microsoft might have done this. The proceedings wandered from analogy to analogy, but none was satisfactory. Holley brought up an AM/FM radio that had the button to switch between bands disabled, but it was unconvincing that what the DoJ was trying to achieve was to force Microsoft to disable IE in a similar way, and Holley failed to make this point anyway. Questions that he had asked the previous day were repeated to no avail. There were long periods when the cross-examination was more of a casual chat than a searching examination, with subjects like Heathkits, FCC regulations, and how Linux (and previously BSD) allowed Farber to explore with his students how the operating system would need to be modified for very high-speed communication. Holley's point eventually seemed to be that Microsoft would have to change its operating system in the near future (undefined) to accommodate all-optical networks. Holley had dug out a statement from Farber that existing operating systems designs would not work with high-speed networks, and that it would become necessary to have lean, mean operating systems - kernel operating systems - and to shorten the paths within them. It would also be necessary to get rid of redundant code that is not used. The result of all this was that Holley won the point that Microsoft would need to continue to develop Windows, but lost the point that bloatware would be tolerable in the future. Unfortunately Farber did not introduce into his replies - and the redirect examination missed it - the point that loading DLLs unnecessarily took a considerable amount of time when a PC is booted, or put into sleep mode in a notebook, quite apart from the RAM real estate that they occupy. Farber complained that when he tried to use Netscape with Windows 98, IE kept appearing. Holley quickly changed the subject. Nor did Holley like the refutation of his suggestion that any brilliant student could commercialise a great software idea: Farber noted that Andreessen "had Jim Clarke next to him with a large pot of money and a lot of intelligence". Holley tried to isolate Farber by suggesting he wanted Microsoft to be denied its copyrights in its products, and have them put in the public domain. Unfortunately Farber backed away from this eminently sensible suggestion and ran up a "I want to pick and choose" flag, with the code that was not chosen remaining in the library. Holley chose to misinterpret this several times, even after Judge Jackson had tried to terminate the silliness, to imply that Microsoft would have to test 1,000 or 10,000 versions of Windows 98, which was of course utterly absurd. Holley claimed that Microsoft had a code tester for every code writer, but Farber missed the opportunity to enquire why there were so many bugs in Microsoft software if this were really true. When Farber again mentioned that he had had to remove Netscape because IE kept popping up when he did not want it, Holley abruptly terminated his cross-examination. The redirect examination was in marked contrast to the earlier examination: it was conducted by Denise de Mory and was precise and to the point. She scored an immediate goal when it was revealed that one of Microsoft's intended witnesses, Professor Michael Dertouzos of MIT, when asked if a browser was an application had testified that "Historically and today, it is the case that browsers are treated as applications." Needless to say, Microsoft announced a substitute. She went on to get a second goal when she produced the Microsoft Press computer dictionary, where IE was described as a web browser, and web browsers were defined as "a client application..." After Holley's efforts the previous day to find definitions of operating systems that suited Microsoft's case, de Mory asked Ferber if he agreed with the definition in Microsoft's computer dictionary. He did, and it destroyed Holley's previous efforts. So far as separating the OS from the browser was concerned, Farber said there were a lot of benefits. De Mory then turned to whether it was necessary for Farber to understand the details of Windows 98 to reach his views: of course it didn't. A further clincher was Farber's opinion that Microsoft could have designed Windows 98 so that IE was separate, but users would have the same benefits. Farber also noted that a choice of browser would allow OEMs to differentiate their offerings more readily. De Mory concluded with another extract from Partovi's evidence, which pointed out that in IE5, Microsoft was moving around functionality from DLL to DLL, so demonstrating the flexibility of packaging available to Microsoft. It seemed that the DoJ won this round on points. ® Complete Register trial coverage

Sponsored: 10 ways wire data helps conquer IT complexity