My thoughts on this, even though it was addressed primarily to Joerg. On Tue, Dec 18, 2012 at 12:09 PM, james <ja...@mansionfamily.plus.com>wrote:
> Let me mention what I am interested in: >> >> Binary compatibility to existing (pre Oracle) Solaris installations. >> > Why? I have software. There are vendors looking at moving from Solaris to a less-corporate-competitive alternative. > I've enjoyed Solaris binary compatibility with database and market data > systems and old apps as much as anyone. But that was in environments that > had paid for those things and were paying for support etc. They will not > be using anything thisgroup will create for that anytime soon. That is purely our fault. We have executed badly. > Compatibility to SVID3 (Svr4 compliance) where possible. >> > Why? > Encourage software adoption by a broader audience (i.e. ISV's) > > Compatibility to all POSIX versions if possible. >> > Why? > Encourage software adoption by a broader audience (i.e. ISV's) > > (Yes I know this is heretical on one level; but what do you get *in > practice* that actually matters?) > Stability for software developers, so they do not have to revisit old code, and allow more time for them to create higher-level value. > > The problem is: by the time you attain standards compatibility, what will > be the relevance of that compatibility for ordinary people, who just want > to do their day job? And how many people will you alienate and > disenfranchise by slavish attempts to stick to de jure standards? > That is why both is done. Core-OS, standards. Legacy packages for compatibility. Bleeding-edge packages for others. PATH does wonders. > A de jure standard is only useful if its what people need. Has it occured > to you that a shift to a (sometimes inferior) de facto standard might > indicate that the de jure standard just isn't that useful any more? > It is all about money, in the end. If you get something close for nothing, standards matter less. > When there was half a dozen big-iron UNIX vendors and we all had > heterogeneous networks, sure - it all mattered. But I'm really not so sure > it does anymore. > Ultimately, they priced themselved out of the market, without enough innovation. Sun did the closest to innovating. > There is a level on which I think Ian was trying to save you from > yourself. Sun certainly went on and executed badly though, no question > about that. > The idea of a network based packaging interface was goodness. By creating ANOTHER packaging defacto-standard, there was little saving done. Ian should have advocated wrapping SVR4 Packaging and created a super-standard. Increasing the number of languages needed to support the "core" was probably not very wise, either. No ISV will want to move to another [proprietary] packaging standard when they are using off-the-shelf packaging tools to support multiple platforms. Of course, I suspect Joerg's thoughts will be entire different from mine. ;-)
_______________________________________________ belenix-dev mailing list belenix-dev@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/belenix-dev
