Jon Lasser wrote:
And from what I've heard about HotJava (not seen it yet, can't comment strongly) there needs another jump in PC power before it would be useful at the home level. Part of why we were stuck with DOS for so long is that it was what got the job done when the revolution happened. For that reason, I agree that HTML/Integrated browser solutions are what we're looking at, and at the same time don't have strong hopes for HotJava, though I would like to see it succeed.
HotJava runs fine on a 486/33 with 16mb running WinNT 3.5. 486s are pretty much standard. 1995 and '96 will be "the year of the pentium" Entry level systems are now Pentium 90s with PCI and 64-bit video. Pentium 120 systems now cost less than my 486/66 system did a year ago. HotJava is compiled into efficient byte-code with the option to be translated to machine code at run time. This is the same principle behind the Newton. Since the majority of the CPU time is spent in native-C code function calls to the user interface, and network latency is high, the "slow" interpreted code is hardly noticed. In fact, I wouldn't call the Java runtime slow, it beats the performance of many Lisp interpreters which have been adequate for "home" users. (e.g. Emacs Lisp) HotJava is not meant for writing applications to decode MPEG in real time. It's best use is for interactivity on web pages. It doesn't take a powerhouse of CPU to put up a slider, "sleep" for an event, and they call a ScrollList() routine. There's going to be a huge use of Java for doing sales catalogs and online ordering on the web.
most PC users like "pretty" better. And they made the best "pretty" software.
Netscape software is not just a pretty user interface, it also has the best layout algorithms I've seen of browsers. That's some pretty hefty dynamic programming there. Mosaic and Arena frequently produce poor or incorrectly formatted pages. (or less optimal pages than I've seen Netscape) -Ray