Hi, > > 5) The disk partitioning functions would be useful on an installed > > system. > > I see three ways to do this: > 1) Import sysinst into base; > 2) Create separate disk management utility and just run it during > installation process; > 3) Write library and use it for sysinst and for new utility. > What way will be selected - is more political issue, I think. 2) might be difficult, as atm the partition usage (i.e. which partition for which filesystem) is assigned before the actual partitioning (I think that's the better way to go, you don't have to write the label before deciding for the usage). With the other capabilities of sysinst, I think 1) might be the best way (e.g. for configuring the network, installing packages, etc.). > > 6) I'm not sure of the value of trying to have too much in the way of > > automated installs - if you don't have identical hardware, or > > aren't doing a very simple 'use all the disk' install they aren't > > that much use. > > Automated install means that user create answers-file to sysinst, write it > to installation media, insert disc (or boot thought PXE) and after 10-15 > minutes get installed system. Also answer file may contains directives to > configure network and start ssh with predefined password/key. That may be > useful for machines without keyboard/monitor or in missions of mass > installation (clusters?). If add binary packages installation and more > post-configure abilities to sysinst we'll got powerful tool. I did this once, but then decided against (and wrote a script to do so). There are several confirmations being asked for, repetitions, etc., as David Laight said, and then "useless" questions you wouldn't answer on an automatic installation... Putting automation into menuc/msgc is much work as well, as sometimes there are actions inside the menus. There were about 60 or 70 questions you had to automate when covering the whole installation process with all its capabilities. Squeezing the automation into sysinst (and especially testing them!) seemed too much effort and too error-prone to me, compared to making another simple script which perhaps asks for a single parameter if it's wrong. But I still think automated installations are very good. For distributing NetBSD among a larger number of servers or especially clients manual installation takes hours. After all, you're right, I think. The usage of sysinst is imho still the best among all the installers I saw so far, and building and maintaining single master tool could be more effort than having separate tools, which then also have improved usability. But maybe thinking about modifications to menuc/msgc to support a line-scrolling cli, or graphical, or cgi output, could also be an idea (of course too much for this gsoc project). Regards, Julian
Attachment:
signature.asc
Description: PGP signature