Peter Tribble wrote: > On Wed, May 27, 2009 at 8:05 AM, Dave Miner <dminer at opensolaris.org> wrote: >> Glenn Lagasse wrote: >> ... >> >> The above represents one of the criteria that have been applied in >> considering tasks available in the OpenSolaris installer. Glenn's pointed >> out one specific example that we haven't implemented yet, but remains under >> consideration. >> >> Another that's specific to interactive installers is to perform the set of >> tasks that "all" of the targeted users require in order to have a >> functioning system at the conclusion of the installation process. I say >> "all" because it's not truly 100%, but the threshold should be a >> super-majority of, say, 80%. This protects the usability of the installer >> for the majority by not inconveniencing them with unusual cases. > > Depends on your targeted users. A common question I see is "how do I > partition my drives and select the software to be installed?". Rather than > simply saying that advanced uses aren't allowed, we should work out how > to present an interface that allows expert use without interrupting the smooth > flow for the common case. > >> Yet another criterion is whether the functionality truly requires user >> input, or instead can be accommodated by changes to the system's default >> behavior, or can be computed or inferred based on other choices. >> >> A key, but often overlooked, phrase in the above is *targeted users*. The >> current GUI, and live CD, is primarily targeted at desktop/laptop users who >> are new to OpenSolaris, because that represents a significant avenue for >> attracting developers. > > There's no doubt that the new installer has some success in that area. I > don't think that the Live CD is really suitable for the targeted developer - > primarily because there are significant things missing off the Live CD (such > as the total lack of any development tools). At the moment the Live CD is > reasonable for NetBook style use (but even there is short of key applications) > and demonstration, but it's a lot more work later to get it to a useful state > for other usage patterns. >
Every installer medium is going to lack something that someone wants. A successful strategy here doesn't focus on the installer, but instead on ensuring that, as needs change over time, the system can be easily adapted to them. That's not to say we can't make the installer better, too, but it is the wrong place to start most of the time. > Also, while it's good to get developers attracted to the platform, we need to > make sure that they stay with the platform through deployment, and we're > doing nothing to help there. > Huh? I reject the notion that putting thousands of packages into a repository that's usually just a click away, rather than requiring one to hunt up a piece of media that will more than likely contain an obsolete version, is "doing nothing". We of course can do *more*, but credit for the point we have already reached is due. Dave
