> On Feb 4, 2008 7:30 PM, Ken Gunderson > <[EMAIL PROTECTED]> wrote: [snip] > > Linux => GPL license. Linux kernel bundled into > assorted "distros" with > > various GNU tools (Gnu's Not Unix, proclaimed loud > and proud on their > > homepage) and disparate userlands, etc., the > components of which are > > far from standardized. Dependency nightmares > during installs and > > upgrades. When several _hundred_ RH boxes are > borked because of sloppy > > merges, etc., who cares that it's commercially > "supported", you still > > have one hell of a lot of unhappy customers yelling > at you. Granted, > > Linux has improved over the years, but still not > without it's warts. > > I fail to see what this has to do with anything that > is being done > with Project Indiana or OpenSOlaris in general. > > Regardless of the problems that these platforms have, > right now, they > are the ones leading the market. >
I'm dubious of the strategy implied by this remark. Trying to gain market share by emulating a market leader is generally an exercise in futility. People don't make large-scale platform migrations unless there is a significant comparative advantage. Do you really think the GNU userland provides Linux with such a comparative advantage? I sincerely doubt it -- most people use bash due to inertia, not because it is inherently "better" than ksh93. If you want compelling reasons to run Linux, look no further than a broad range of device drivers and sane package management. Both of these are being addressed without any dependency on moving towards a GNU userspace. I think a large part of the argument for a GNU userland revolves around the false premise that "approachability" and "ease of platform migration" are synonymous. > > Which brings us to Solaris. Can Solaris provide > integrated, cohesive > > kernel and true unix userland in a stable and well > performing package > > that is freely available, and hence able to compete > on it's own > > technical merits with the freely available *BSD's > and Linuxes? I hope > > I find the phrase "true unix userland" to be rather > funny. Solaris > ships with several flavours of "userland" utilities, > many of which > aren't even in the default path configuration. > > Most of the utilities in the default path (such as > grep, etc.) are > widely panned for the lack of modern functionality, > updates, and > numerous bugs. > > While I personally I am very grateful for the xpg4, > xpg6, and other > standards compliant environments that are available; > there is no > reason that the default userland has to resemble the > olden days of the > pdp-11 :) (joking) > If you consider the standards to be outdated, then the appropriate course of action is to push for ratification of new revisions to the standards. Adoption of the GNU toolchain simply perpetuates the widely-held belief that standards-driven development is dead. That's hardly the sort of example a project which purports to be best-of-breed should be setting. > While I wish that there were better options than GNU > userland in terms > of functionality, the reality is that the majority of > the open source > world has chosen the GNU toolset, for better or > worse, as their > toolset of choice. > I vehemently disagree with the notion that popularity should dictate product engineering. Users are fickle. Popularity can be quite fleeting. Strategic decisions should not be based on the current direction of the wind. There are many cases where emancipation will take longer, but prove strategically beneficial. > Wasting precious resources on attempting to reinvent > the GNU wheel, > all in the name of (mostly) pride and arrogance, > isn't going to win > any battles. > Pride and arrogance have absolutely nothing to do with the large-scale opposition to adoption of the GNU userland. Unless by 'arrogance', you mean the belief that software should be engineered instead of written in an ad-hoc "organic" manner. I'm far more concerned by this (frankly inexplicable) haste to get to the promised land of an approachable desktop-oriented Solaris distribution, than I am about the actual implementation details. There is a broad perception that the core design principles which have served to differentiate Solaris from linux and its ilk seem to have largely been forgotten. I cannot stress enough that this perception is largely fueled by the fervor with which a few individuals promote change purely for the sake of change. I have absolutely no problem with the project Indiana concept. To the contrary, I'm quite happy that such an endeavour is being undertaken. However, the strategy of supplanting vast amounts of code with the GNU toolchain is a suboptimal path of least resistance. Emancipation is hard. GNU user re-education is hard. So what? Hard in and of itself should never serve as justification for not doing something. > Ivory tower attitudes will keep that tower sparkling > white, so that > future generations may remember it fondly as they > migrate to systems > that actually meet their needs. > Uh, nice straw man. I'm not aware of anyone who actually thinks the status quo is acceptable. Rather, there is a contingent who don't fit within your (false) dichotomy -- they believe the path towards a more approachable product is best taken through thoughtful architectural review and emancipation. Will this take time? Absolutely. Is it time well spent? I'd argue that it is to our strategic advantage to take the time. If Indiana is to value backwards compatibility even mildly as much as Solaris, then getting things right the first time around will be absolutely crucial. > > so. That's why I'm looking at it in the first > place. As a professional > > unix sysadmin I'm not too interested in yet another > "Linux distro > > of the month" to play with nights and weekends > because I have no other > > life. So what's the Solaris target market going to > be, professionals or > > hobbyists? There's lots more of the latter if > you're objective is > > mindshare with the pc hobbyist rags, etc., wh/may > do quite well at > > raising visibility. But I don't think these folks > buy support > > contracts, nor are they likely to upgrade to Sun > "big iron" sparc > > machines. > > I'm not interested in a yet another "Linux distro of > the month" > either; that's why I'm excited about things like: > IPS, the > Distribution Constructor, Caiman, and others. All of > these projects > are taking fairly different approaches to the same > problems others > have tried to solve before. IPS, notably, stands out > the most as being > widely divergent in its approach to packaging. > > I would encourage you to approach the efforts of this > community with > an open mind. > Many of us are quite open to change. However, we're also not naive. Rapid change is a perfect vehicle for poorly reasoned schemes to slip through the cracks. Many of the things which will come out of Indiana will be extremely beneficial to OpenSolaris. However, these developments should not serve to absolve Indiana for any future misdeeds. > I do not believe for one moment that the engineers at Sun will replace > anything in Solaris without careful consideration and planning. Nor do I. Sun stands too much to lose from the alienation of their old users. I'm confident that cooler heads will prevail. -Tom This message posted from opensolaris.org _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
