>>>>> "g" == Gary <[email protected]> writes: >>>>> "m" == Michael <[email protected]> writes:
g> -- and perhaps why it didn't get much traction all around --
g> is that there are probably not nearly as many closed source,
g> proprietary Linux apps as there are closed source, proprietary
g> Solaris, HP-UX, Irix, Ultrix, SCO, AIX, etc.
maybe, but don't you think that the fact that it ran only very old
distributions, and that it wasn't good enough to run Apache, might
have something to do with it?
g> So unless this project
g> attracts that kind of man hour backing, I'll have to pitch my
g> vote with the others in suggesting it would be nice but not at
g> the expense of core functionality
IMHO it would have been wise to ditch all work on native X servers and
JDS, and put it into lx brand instead: use actual Linux binaries to
implement the desktop, including even the X server, instead of
rebuilding old crappy versions of them with all the logos substituted.
It could end up being less work for much bigger payoff. ...though the
X server part may be difficult to impossible since the proprietary
Linux servers come with an enormous kernel module blob.
This kind of light-weight virtualization could have been good for
desktop users because now the ONLY way to run a desktop is inside a
zone, so why not have more than one zone, in particular for the sake
of having separate IP stacks. You could have ``work'' and
``personal'' zones, one on the VPN and one not. or if you live
outside US you might have a local-transit zone to keep local CDN
access or if you have an .au-style intl data cap, and also a
tunneled-to-US zone for using US-based DRM crappo like pandora, hulu,
whatever through one of those $5/mo VPN companies.
It leaves the solaris userland free to worry about
performance-optimising a few key web applications like nginx or
Postgres that might benefit from tweaks to posix threading or
epoll/kevent/eventport, integrate HA auto-start-stop features, and
make backroom deals with big-proprietary-blob apps like iPlanet or
Oracle DB that want coordinated release-engineering and logo-certified
cross-promotion baloney. Building a desktop to compete with Linux is
just doomed: they were unable to keep Firefox patched, and couldn't
deliver Chrome at all, and pissed away all their time trying to play
audio (OMGwow), automount USB sticks or build two-years-old versions
of FLASH for sparc! wtf? I understand the idea of eating your own
dogfood, but when your food tastes like stale shit all the time maybe
it's better to rearchitect things.
but yeah it is OT dreaming since there is no one left to do the work.
m> another thing to consider is that the linux kernel the lx
m> brand would need to emulate is a moving target ...
the target it actually did emulate, was Linux 2.4, and yet it was
still useful enough to ``run acrobat'' as you keep saying. probably
not chrome. but...
the actual delivered kernel API of Linux is, by practical observation,
more stable than that of OpenSolaris, in that you can run particular
proprietary apps as well as entire Linux userlands from releases that
span several years under the same lx brand, but to run an OpenSolaris
ipkg zone your build has to match exactly and proprietary apps are
obsessed about completely silly things like whether you have ksh93 or
not.
pgpiospZsqBsQ.pgp
Description: PGP signature
_______________________________________________ Discuss mailing list [email protected] http://lists.illumos.org/m/listinfo/discuss
