alternative theory: the different impact levels were produced by (1) multiple signal strengths or (2) multiple signal durations, either of which would cause different RF emissions for the different characters, and "sound" was just his way of tipping them off to the key problem.
if sound was really part of it, you'd have to equalize the amount of time the wheel spun for each character (otherwise the time between strikes would give you information about the characters used) and the strike area of each character (otherwise the sound level would give you information). hmm - now that i think about it, the timing attack would work for RF, too...
Relational Technology (RTI, subsequently Ingres Corp., then part of ASK, now part of CAI) was the commercialization of the University INGRES codebase. INGRES was a Berkeley project from the early 70s to the mid 80s.
Miro Systems (subsequently Montage Software, then Illustra Information Technologies, now part of Informix) was the commercialization of the University POSTGRES codebase. POSTGRES was a Berkeley project from the mid 80s to the mid 90s.
stonebraker has always released his research software in source form, even in the 70s and early 80s when this wasn't particularly fashionable (these days, the funding agencies make a much bigger deal of it). external contributors were always able to donate changes which were folded into the postgres source tree at berkeley. (most ports happened this way.) the only real change that happened with regards to licensing is that there used to be a UC-imposed license fee for commercial use. a couple of grad students talked stonebraker into talking UC into dropping this fee.
systems like illustra have come to be known as ORDBMS (object-relational), which is very different from OODBMS. don't expect OQL - expect SQL3/SQL-1999.
with respect to printing out web pages, i don't understand why people are so quick to characterize other folks as being scared of "the computer" or "the internet." i think everyone's been burned by all of the following problems:
you can't find a page again.
you can't bookmark a page because the site uses dynamic content in a way that is bookmark-hostile.
you bookmark something, but the site organization changes and the bookmark is useless.
basically, printing is the only option most mortals have for making a persistent note of web content. "save as text/source" or "send page" don't work well because the layout, images and other graphical information are lost. "save as " burns a lot of disk space and introduces a big file management problem.
Slashdot Mirror
if sound was really part of it, you'd have to equalize the amount of time the wheel spun for each character (otherwise the time between strikes would give you information about the characters used) and the strike area of each character (otherwise the sound level would give you information). hmm - now that i think about it, the timing attack would work for RF, too...
basically, printing is the only option most mortals have for making a persistent note of web content. "save as text/source" or "send page" don't work well because the layout, images and other graphical information are lost. "save as " burns a lot of disk space and introduces a big file management problem.
for information on one such project (Xerox PARC's Electronic Paper), see http://www.parc.xerox.com/dhl/projec ts/epaper/.