> >If there is no money in it then there is no business > case for it. Plain and > >simple. Open source is not all about some utopian > vision of free wheeling > >and fancy free give aways. It is business. > >- > >Dennis Clarke > > Huh?? Not everything in life is devoted to the > purpose of making money or getting rich. Every > business case is not written with the purpose of > making money. Business is business - but a business > operating on the purposes of generating money > **only** is a bit foolhearty in the long run. I'd > think you'd want to provide a service or product to > the community as one of those top goals. So I'd think > we'd talk about the NEED or WANT of such a product or > service as one of our top discussions.
I think people needn't be saying different things here. For a business, in the long run, there needs to be a way to show how something adds (money) to the bottom line. But in the short run, it can be about an expanding market presence, goodwill, or something other than boosting the figures in the quarterly report. > Something like: > 1. Does anyone, on planet Earth, need or want an > OpenSolaris-based PowerPC port?!? Obviously, or nobody would be here. > 2. Can the port be marketed and sold commercially or > academically?!? Hmm...I've seen people talk about things they'd like to do or systems they'd like to run it on, but I haven't seen anyone talk here about how it could be marketed and sold. Not that they would, such things typically aren't discussed publically before they're ready for an actual announcement (i.e. a plan and a fairly solid date are already in place). Maybe I missed it though... > 3. Can we utilize existing sponsorship (Genesi) or > seek out new sponsorship (IBM/Sun/Blastwave/etc..) > with interest in our idea(s)?? IIRC, Dennis has said he was looking into that, about as much as anyone outside of Sun could, anyway. > uestion #1 should take precedence in this case. Needs > and wants help create DEMAND. Without demand for the > product or service (i.e. Polaris or OpenSolaris/PPC > (/POWER)), the business case or venture is eventually > short-lived whether through sponsorship or > 'out-of-pocket' expenses. > > In theory, it only takes one person or business > (internal/external) to create a demand... ;o) Yeah, maybe that's how the old 2.5.1 port happened, but it got orphaned pretty quick. You want something to last, it probably takes more than that. Heck, given that Linux now sort of runs on the Wii, I'd love to see OpenSolaris on that, since I have one (and the WiiRemotes are way cool). But I doubt there'd be enough people to make it happen, and anyway it'd be a hack 'cause it takes one just to break into where one can load arbitrary code. Serious desktops, still in production, that use PowerPC? I don't know of anything other than IBM offhand. They probably account for most of the PowerPC or POWER servers, too. Lots of embedded, special purpose, and game systems, but you still need a market, and a development box with observability, and a place to get the chipset drivers going. Looks to me like either get IBM to cough up specs and work with them, or go for the little stuff (which will be even more architecturally diverse and thus difficult than x86 platforms IMO). >From here on, I'm just rambling... One thing I like about any port that has some hope: it's an insurance policy. The revival and revitalization of x86 meant opportunities for selling quality x86 boxes, and while the T1, T2, and eventually Rock should keep SPARC viable for awhile, it's always interesting to watch Sun try to keep up with folks with the $$ to throw around of Intel. Solaris on x86 means that Solaris doesn't depend on SPARC alone. So any port, even if it doesn't become commercially viable, should, if the process is well documented, make future ports a little easier, improve modularity and code isolation, point out more hidden assumptions, etc. Although a lot probably is covered by the SPARC vs x86, given different byte order; even more so with x86 still supporting both 32 and 64 bit kernels. And the Solaris code is from what I've seen a heck of a lot cleaner than some I remember, like SVR2, and way better than the ancient U If one day blade servers were standardized enough that different vendor's blades could plug into the same chassis, one might have IBM blades plugged into Sun chassis, or vice versa. Maybe under those conditions there'd also be interest in additional OS vs platform combinations, although there'd still be the problem of reaching a critical volume sufficient to get application vendors to support those combinations. This message posted from opensolaris.org
