Karen Tung wrote: > Keith Mitchell wrote: >>> >>> In my opinion, the text installer should: >>> >>> - work on both x86 and sparc >>> - offer a similar installation experience and feature set as the >>> existing slim GUI installer >> Should it offer a similar installation experience? What does this >> mean? Is the target audience for the text installer the same as the >> target audience for the GUI installer (I would say that it's not). In >> that case, the experience - and feature set - should be tailored to >> the text installer's anticipated user base. > For me, a "similar installation experience" means similar questions will > be asked. Same things > can be configured/selected during installation. Not more, not less. > > I agree that the text installer's anticipated user base should be > considered. So, let's talk about that. > From what I can see, users that can benefit from using the text > installer are: > > - all sparc users, since AI is the only way to install SPARC machine at > this time. > - All x86 users who can not use the GUI installer now, because their > machine doesn't have enough > memory to run GNOME, or because their machine can't use the window > system..etc.. >
I think the most interesting use case here for x86 is servers which have only a serial-type console available. > Based on the above user groups, I don't see why we need a different > feature set for > the text installer. Offering the same feature set would provide the > same installation experience, > and we can re-use a lot of existing code. > I'd suggest that we have some internal data regarding server customer requirements which I've passed along a while ago, and that should be considered for requirements, even if they ultimately prove to be beyond the scope of the project phase covered here. >>> - the installation performance shouldn't defer too much from the GUI >>> installer. (ie: if it >>> takes 20 minutes to install via GUI, it should take about the same >>> amount of time to install >>> via the text installer) >> Can we define "same amount of time"? Would that be within 10%? No more >> than twice as long? Does installation time include the time the user >> spends selecting options, or just the time spent moving bits? > I guess this point is kinda covered in the "user experience" point > above. I think the questions to be > answered should be similar and the amount of time spent on moving bits > should be similar. > This is assuming we provide the same feature set for both installers. > If they have different feature > sets, probably doesn't make sense to impose this requirement. > Defining an expected performance level seems to be useful for any feature set. Dave
