Thanks for following up, Andrew, a bit more below: Andrew J Guay wrote: ... > What I'm suggesting and suggesting without any clear understanding of > the scope & or difficulty of creating such a thing. A New Tool that > can detect the language a program is written in, the hardware > necessary for running the program, & install the program into cross > platform Operating Systems. Linux/Unix and I say them both from an > open source stand point need to fix the accessibility of the vast > array of software available. Dependencies, lack of standards, > difficulty of installation etc are preventing Open Software from > being widely accepted. You Meaning Sun have cross platformed the > OpenOffice Program it is great and as far as I can see is the best. I > say go one step further be creative do things that haven't been done > before. As I have stated before Sun Knows the different platforms and > can if anyone can accomplish such a thing and do more in this area. I > believe that if your OS is going to compete it will have to stand > out. It Now does, due to world records in performance don't stop > bring the mastery that Sun is Capable of to bear in the market place. > This can be done by making yours the best. I would agree that you > need not do more than meet the standard of today with ease of > installation but that is a big undertaking based on what you have > stated. My Idea is intended to get you to think not just about > installation but utilization once your OS becomes installable for the > masses.
Yes, software availability on Solaris is certainly a requirement in making headway in any market - hardly anyone buys an OS just to run an OS, they buy it to run an application, so the application they need has to be on our platform. We have a lot of ways of attacking that problem. Cross-platform installation technology such as you're describing has been talked about before; back in the early 90's the OSF/DME software distribution effort had the goal of shipping software in some intermediate, platform-neutral form that would then be converted to native binaries at installation. I don't think it ever got off the ground (at least that I can recall). There have also been (and continue to be) emulation environments which attempted to solve that problem, and bytecode platforms such as Java embody some of the same attributes. Overall, it feels like a much bigger problem than installation, so I have to consider it out of scope for this strategy. ... > I can appreciate the complexity of having to create the drivers. Sad > to say for Solaris if the Open Source bank cannot be utilized this > could be it's demise. Thus the need to create tools that can detect & > possibly create what is needed on the fly. I'm not sure even how a > driver is created so I'm sure I speak from ignorance but it seems to > me that lets say a DVD- ROM with an interactive firmware must > communicate with the OS in some manner I'm wondering if their is a > way to utilize these communications to create what is needed. A sort > of automated creation, to eliminate the intensity of hand creating > each and every driver via backwards engineering or what have you. > Imagine how popular Solaris would be if hardware manufacturers didn't > have to create drivers & could simply create a more interactive > firmware & Solaris could automatically create the drivers. > Unfortunately, the details of writing drivers mean that they're very rarely easily portable across platforms, and certainly not automatically. In areas such as networking, technologies like ndiswrapper allow for some Windows drivers to be used on other platforms. But usually those perform poorly and don't work as reliably as users demand. Perhaps someday, but it's very far in the future. Dave
