oster@lapis.berkeley.edu.UUCP (02/10/87)
Take that paper with a boulder of salt. It concluded that X was better than NEWS using the following (idiotic) argument: 1.) NEWS supports an extensible communications protocol, in which an application can redefine the protocol it uses to talk to the display server to make the protocol more tightly tuned to the applications needs. 2.) SInce you CAN do this in NEWS, the author of the paper assumes your MUST do this. 3.) You cannot do it in X. (The prootocol is low level, and has no extension language.) 4.) The author shows that there are situations where it is not appropriate to extend the communications protocol. 5.) The author concludes that NEWS is bad. This conclusion would hold if (4) were common and (2) were true. Neither is the case. I'm really excited about using NEWS for dynamic load balancing: a program can detect the relative speed of the backend processors and the display processor (which is different each time the program is run because of multi-tasking load.) and dynamically, during the run, move parts of itself between the two processors in order to get the best performance. NEWS can't quite do this, but it is a step closer to having what you need to do this than X is. --- David Phillip Oster -- "The goal of Computer Science is to Arpa: oster@lapis.berkeley.edu -- build something that will last at Uucp: ucbvax!ucblapis!oster -- least until we've finished building it."
RWS@ZERMATT.LCS.MIT.EDU.UUCP (02/10/87)
From: jade!lapis.berkeley.edu!oster@ucbvax.Berkeley.EDU (David Phillip Oster) ... using the following (idiotic) argument: Now, now, let's be polite... 2.) SInce you CAN do this in NEWS, the author of the paper assumes your MUST do this. I never said this. 3.) You cannot do it in X. (The prootocol is low level, and has no extension language.) X is extensible, and one could certainly define a (reasonable) extension language if it seemed desirable. 4.) The author shows that there are situations where it is not appropriate to extend the communications protocol. True enough. 5.) The author concludes that NEWS is bad. This conclusion would hold if (4) were common and (2) were true. Neither is the case. My conclusions were based on more than just the extension language. Good of you to ignore that fact. Also, I assume you have a laundry list of cases where an extension language is useful; perhaps you could publish them, and enlighten us. Perhaps you can tell us you have actually implemented them (both ways), and convince us as well as enlighten us. NEWS can't quite do [dynamic load balancing], I'm glad you said this. but it is a step closer to having what you need to do this than X is. Again, the fact that the X core has no extension language does not preclude defining one.
carlson@styx.UUCP (02/10/87)
In article <2479@jade.BERKELEY.EDU> oster@lapis.berkeley.edu.UUCP (David Phillip Oster) writes: >Take that paper with a boulder of salt. It concluded that X was better >than NEWS using the following (idiotic) argument: > [...etc] Seems to be a lack of understanding on both sides of the sandbox... BTW, I have heard of a paper listing the advantages of NeWS over X. Can someone send me a copy? Does it exist? John Carlson ARPA: carlson@lll-tis-b.arpa UUCP: lll-crg!styx!carlson