rws@EXPO.LCS.MIT.EDU (05/24/89)
There were headlines a couple of weeks ago about "Quarterdeck Patents Windows". I suspect I'm not the only one who was curious. I have a copy of the patent (thanks to Karen Kramaric at the MIT Technology Licensing Office). It was apparently first filed by Quarterdeck Office Systems on May 2, 1984, and then a revised version was filed on March 17, 1988. The Inventor is Gary W. Pope. The patent deals with a particular mechanism for having (multiple) application programs avoid the overhead of making "operating environment calls" (read "system calls") to write into the video display. Each application is given a "Pseudo Screen Buffer" that it can write into as direct memory without interference. Using a timer interrupt, the system periodically compares these buffers against a "Previous Image Buffer", which contains a copy of the actual screen contents. Where they differ, an update is done to the Previous Image and the display. An algorithm for making this "efficient", based on various heuristics, is described. Overlapping windows are handled with a Screen Map: each cell of the Screen Map contains the id of the top-most window overlaying that cell, and updates to a particular cell in the Previous Image and display are conditional on the Pseudo Screen Buffer id matching the id in the Screen Map. Overlap support is clearly assuming character-cell graphics; some of the patent hints at "bits" in addition to characters, but most of the language in the patent deals with "character" output. There is also some stuff about calculating memory locations "cleverly" to make some computations efficient, which appears to be dependent on a very specific memory layout. I can't see anything in the patent that should cause anyone I know to worry.
raveling@venera.isi.edu (Paul Raveling) (05/24/89)
In article <8905241219.AA00789@expire.lcs.mit.edu> rws@EXPO.LCS.MIT.EDU writes: > [About Quarterdeck's "Windows" Patent application, > details omitted] The description sounds remarkably similar to the technique we used in 1968 at UCLA to support the "STAT" function. This was a system operations interface on the URSA "console system" at UCLA, used mainly on the 360/91. At the top level of description the only apparent difference is that STAT didn't handle overlapping windows -- it only did some minimal tiling. Other than that, everything sounds like ideas that are hardly novel. If UCLA still has listings or code it's possible to document that they're 2 decades old. ---------------- Paul Raveling Raveling@isi.edu