Alan DuBoff wrote:
What I am suggesting is that we should finish the work started in the
emancipation community, pull in the groundbreaking build and boot
efforts from the various other distros (starting with Schillix,
Belinix and the IBM/z port simply because Joerg, Moinak and Neale are
involved) and produce the tools, docs and code needed to enable others
to easily grab the OpenSolaris source code from OpenSolaris.org's
mercurial repos, build the bits and boot a running OpenSolaris kernel
without reference to or use of closed-bin bits, proprietary-to-Oracle
binary repositories or tools.
In my mind, *not* being able to do this simple thing ourselves is a
liability that needs to be addressed...
-John
John,
I couldn't agree more, but am in wait-see mode.
I installed Ubuntu on a laptop this week, everything works and is built with
open source. Even the newest OpenOffice with the Oracle logo on it is included
in the latest 10.04 release. VPN (IPSEC/IKE), audio, video, NIC, wifi,
hibernate, sleep, suspend, everything works, I'm using it on a docking station
closed but attached to a monitor. While the Broadcom driver for wifi doesn't
ship with the distro, there is an automated process to strip the firmware out
of the chipset so that the driver can be built.
I am keeping an eye on this community, as I do still have Solaris running on my
2 servers at home, but it seems that Linux will ultimately be more important
for my work.
Amazing that all the codecs seem available that I need, the plugins work, and I
even have a Chrome browser on Linux. Solaris management was always saying that
couldn't be done, but the codecs are available over the wire, transparent in
the distros of Linux. This is how Solaris should be
Be *very* careful what you wish for. One of the problems with
OpenSolaris (from a PR standpoint) has been the lack of real emphasis on
the GOAL for OpenSolaris. Is it a general-purpose, use-it-for-anything
OS? Is it for embeded systems? Backoffice server-room use only?
Desktops? Laptops? Micro-computing (e.g. cellphones)? What targets are
we aiming for?
All of these entail some level of tradeoffs as to what to expect,
feature-wise. You can't have it all with a finite amount of resources to
spent on things.
Solaris used to be pretty obviously targeted at backoffice/server usage,
with tie-ins to corporate managed-desktops (e.g. SunRay). OpenSolaris
seems to have fuzzed this quite a bit, and I think we're in the
situation where people are bound to be disappointed with OpenSolaris,
because there hasn't been a real emphasis on What Is It Really For, and
we've gotten distracted by adding a bunch of random features, fairly
haphazard.
While you may have found Ubuntu 10.04 good for your laptop, I have found
it to be absolute crap for servers. We all need to remember that there
are specialized tools for a reason, and trying to make everything a
hammer/wrench/screwdriver/corkscrew/whatever supertool is bound to
fail. Pick our poison and stick to it.
That said, I'd prefer the OpenSolaris situation to emulate the OpenJDK
situation, where there's a open, community-managed process around a
buildable free source base, but still places where Oracle is able to
value-add for its own purposes. I suspect that was where we were going
until the merger, and I know the Solaris codebase is much harder to free
than the JDK sourcebase was. And, honestly, it's really hard to just
break off from the very elaborate build/test setups Oracle has for
OpenSolaris - replicating that kind of integration process and testing
is hugely expensive, in time, manpower, and cash. Having a "free" tree
somewhere is really of little use without the build/test/codemanagement
system behind it.
Adobe doesn't provide a SPARC release of AcroRead anymore, and Cisco doesn't
provide VPN clients for Solaris on x86 (SPARC only). This quagmire will play
havoc for Oracle, IMO, as well as the community in general. Oracle has deep
pockets to rectify that problem, but time will tell if they feel compelled to
assist.
Having a better focus will lead to some improvements in this area, as
decisions are made about where to put effort, and dump unfruitful
avenues. Sun was famously crappy at that process. Oracle has a better
reputation, but who knows....
--
Erik Trimble
Java System Support
Mailstop: usca22-123
Phone: x17195
Santa Clara, CA
Timezone: US/Pacific (GMT-0800)
_______________________________________________
opensolaris-discuss mailing list
opensolaris-discuss@opensolaris.org
