mwm@pa.dec.com (Mike (My Watch Has Windows) Meyer) (06/05/91)
Masataka Ohta has been making various largely unsupported claims about shared libraries, and doing so in a generally obnoxious manner. Anyone who's been watching USENet for any length of time can predict the results of that - a long, heated discussion about those claims, gradually degenerating from a discussion into an argument and winding up as a shouting match. The last post I saw from Ohta took us to that level. Another predictable result isn't so obvious because its part of the bias of USENet. The people participating in the discussion use examples drawn from Unix systems. This means the implementation of the shared libraries, the interfaces in them, and the system they are running are all suffering from a handicap: they are patches to an existing system, and not designed in from the day 1. Unfortunately, that belief is like the belief that parity is for farmers, that you use real memory for real performance, or that shared libraries are of no benefit, in that the merits of the argument are immaterial. Workstation consumers/vendors have already settled the issue. Reality in the workstation market is simple: Workstaion consumers demand Y (can't use X as a free variable anymore). Workstation vendors perforce provide Y, or lose market share. Chip vendors perforce provide chips that allow Y, or lose market share. This statement is true if Y is Unix or demand paged virtual memory. It's becoming true for shared libraries. While Ohata may think the benefits of shared libraries aren't real, lots of people who buy workstations think they are. Fortunately for Ohata, he can relink all his applications statically and not have to deal with them, yet still get the benefit of buying boxes from the mainstream workstation market. <mike -- Don't tell me how to live my life Mike Meyer Don't tell me what to do mwm@pa.dec.com Repression is always brought about decwrl!mwm by people with politics and attitudes like you.