To: Thomas A Watson who wrote (2355 ) 10/29/2000 11:49:30 AM From: E. Charters Read Replies (1) | Respond to of 2615 Run these: Xpaint --> paint a box and fill with colours. Netscape --> load a few sites. Gimp -> process a photograph for printing Ghostview --> look at a Post Script file Xpdf ---> look at a pdf file Workman ->> -play a song on a CD Applixware ---> load a word document with gifs, of 20 pages ALL AT THE SAME TIME! What happens? gimp and xpaint will interfere. applix will not load. Netscape will go grey and lose text, workman will lose its colour and not load the CD and colours will flash all over the screen. You will not know what colours you loaded or filled in gimp and xpaint. Nothing will work. You will have to reboot X as no program will be closable. In fact, and there is the "piece de resistance", you may have to reboot your whole machine, as the killer with X is that it can lock your whole system. That is why every text available says: DON'T RUN A SERVER on X! All because of the basic insecurity of X. Its applications run their own colour and memory and windows management. How can they know what the other program is doing? They can't all they can do is crash into it. It is not a problem of too little memory in RAM. It is a problem of the OS not handling what memory there is properly. Wrong paradigm. Windowing was broken in X ten years ago and it still is. It is much too complex to build windows with basic OS calls in X. That is why so few people PORT IMPORTANT PROGRAMS to X. THAT IS WHY the BERLIN and GGI projects exist. THAT IS WHY JAVA was developed believe it or not! Not that it solves the problem. People who have written important commercial programs have told me that they set out to port their program to X and it was too hard. Like widely sold accounting programs we have all heard about. It took Corel 8 months and 300 programmer to port Word Perfect to X. Where is that at? If you develop a program in X it has to be tested for conflict with every other X program. If there are 10,000 programs under X how long will testing take if 20 different modules, each with different data on different machines is tested? Let's take the program modules alone. That is 10,000 X 10,000 20 times. That is this many iterations. 100,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000,000 Will automated machine testing help? Perhaps I am exaggerating the complexity of the problem. Perhaps not. You like simple. So do I. If simple works let's keep doing it. But X, while a GUI, and while it does some things a GUI should do, is at the end of its life cycle, and an old, complex life cycle at that. If it was so darn good a lot of people would have jumped to Linux by now. Instead they wrote Gnome and KDE and Enlightenment to "improve" X. And most agree there is not much improvement. Let me ask you this. Why didn't people port X to DOS and run DOS under it and port all the windows programs to X? Why do VMware and WINE exist? Because people's programs that they NEED, do not run under X, and probably never will. EC<:-}netlinux.dynip.com