Brilliant and very useful for FreeBSD, IMHO
Gary W. Swearingen
swear at attbi.com
Tue Apr 8 10:36:26 PDT 2003
Johnson David <DavidJohnson at Siemens.com> writes:
> Frankly, the expections set forth by the reviewer are unrealistic. She
True, but with few exceptions, it's only unrealistic because the many
people with the moderate skills needed to do the job are unwilling to
invest their time on the job. (I include myself in this category.) I'm
not sure why this is. The most skilled and ambitious seem to prefer
hacking the guts of the OS or leading their own application project. I
suspect that the job of improving the User Experience requires too much
difficult human interaction by designers, etc.
I've observed a couple of Linux distribution developments which make
me think that the problem of "ease of use" is more one of desire to do
the job well than one of huge difficulty/manpower.
The first is "Red Hat Linux" from 1995-2000 while I used it. Until
fairly late in that period, almost all of the installation stuff was
developed by one or two guys and almost all of it was developed early
on. Horrible, poorly-documented, user-unfriendly crap that they just
kept pushing out the door while improving their BS marketing machine.
They spent millions on lord knows what, supporting kernel developers,
GNOME people to re-develop the GUI wheel to replace KDE, etc. They
could have easily afforded to polish their installer using the same old
character-based GUI, or even the development of a real GUI installer
like their poor cousin Mandrake did. Eventually, they made a poor
attempt to catch up with Mandrake, but it's still unfriendly, even
(and especially?) for experts.
The second is "Lycoris" (formerly "Redmond Linux"). This was mostly
developed by one guy (with some help from a few remote helpers). I
laughed when I heard his early ambitions, but I've read at least one
glowing review of his product and it has satisfied at least one OEM.
> wants a Windows clone.
If an easy-to-install GUI Linux OS is a Windows clone, then yes. Few
people wouldn't want such an OS.
Even without better automating the install process, the FreeBSD
installer could be greatly improved with *relatively* little effort.
The thing is just un-polished and unfriendly. Several confusingly-
different ways of navigating menus; use of esoteric, undocumented terms,
poor help system, poor explanations of what's happening at each step,
etc. The many efforts devoted to a developing better GUI installer
software would have been better spent re-thinking the use of the old VGA
software, but of course, that wouldn't have been as much fun and one
can't fault people for doing what pleases them instead of doing what
would be better for FreeBSD.
I might as well mention my pipe-dream installer. It's text-based; has
few features and fewer options; boots off one or a few floppies (or
CDROM); offers to install to pre-existing unpartitioned areas of hard
disks with selectable levels of "autoness"; offers to partition disks;
installs only a minimial OS (ideally, text or VGA option) to a temporary
OS directory tree from which the real OS installation (selectable binary
or source, from CDROM, FTP, NFS, etc. distribution) is performed. The
temporary OS has sufficient tools for allowing OS config tweaks, FreeBSD
Handbook & FAQ, and a good software installer (same used for install and
updates of both kernel and basic & ported applications) which allows the
real OS and applications to be selected (by pre-configured sets and/or
individually). Ideally, it would come with M$Win software to allow the
user to use M$Win tools to pack the M$Win partitions and create
unpartitioned areas of disk for dual booters who want to share a disk.
More information about the freebsd-advocacy