[OpenIndiana-discuss] "OpenIndiana lead Alasdair Lumsden resigns"

Reginald Beardsley pulaskite at yahoo.com
Sat Sep 1 14:19:58 UTC 2012


--- On Sat, 9/1/12, Robin Axelsson <gu99roax at student.chalmers.se> wrote:

> From: Robin Axelsson <gu99roax at student.chalmers.se>
> Subject: Re: [OpenIndiana-discuss] "OpenIndiana lead Alasdair Lumsden resigns"
> To: openindiana-discuss at openindiana.org
> Date: Saturday, September 1, 2012, 7:27 AM
> While I don't have a clue about what
> userbase OpenIndiana has and how widespread it is, there are
> some things I see that don't look good for OI.
> 
> First of all I find it to be poorly marketed. The website is
> updated almost never and it looks like nothing is happening,
> there are no roadmaps, the documentation on the site is
> improving but still has some considerable lacks. A website
> that looks poorly maintained with empty menus doesn't look
> good at all. The website should put a much greater effort at
> marketing itself and show what OI/Illumos can really do.
> We're talking about statements such as "ZFS is leading
> technology ..." and rich illustrations so that even less
> versed people will understand it. The design of a website
> also communicate the quality of the product. It may sound a
> little vainglorious to some people but that's how it works
> in real life.
> 

It's quite a lot of work to do what you're asking for.  Commercial companies (e.g. IBM, RH, Suse) are spending a lot of money on staff to fund this on Linux.  The corresponding source of funding for OpenSolaris was Sun.  OI came into existence when Oracle withdrew support for OpenSolaris and closed the code.
The remaining players in the OpenSolaris community don't have the deep pockets needed to do everything and must choose carefully what they do.  The rest is volunteers.  If you see something you think needs improvement work on it.

> Secondly I find it incredibly hard to start developing
> things on the operating system. If I for example want to get
> started with contributing by porting a Linux graphics driver
> to OI/Illumos, where do I begin? How do I get the compilers
> to work with me without errors and how do I troubleshoot
> them? Or in short ..... where is the *documentation* for it?
> I think there should be an open door that lets new people in
> and makes it easier to get started developing for
> OI/Illumos. Right now it looks like a closed community with
> a very high barrier of entry for those outside that are
> willing to develop for OI/Illumos. If this problem gets
> fixed then maybe userland applications that are necessary
> for a desktop OS will eventually find their way into the OS
> for those people who want to use it as a desktop OS.

There is a small mountain of documentation. Whether it's still as accessible as it used to be I don't know.  I assembled a local copy of all the Sun docs long ago.  

There is inevitably a high barrier to entry for someone who wants to port a device driver from one OS to another.  It requires a good understanding of two complex systems and the physical device.  That is never going to be easy.  In many instances it may be easier to write a new driver from scratch.

As for developing on Solaris generally, it's not noticeably different from any other Unix derived system other than being a good bit more polished.  This not quite as true for OI, but it's hard for me to understand what issues you're having in that area. It is not the same as Gnu/Linux, but that is because tradition and compatibility are given far higher precedence in Solaris.  The big objection long time SunOS users have to Gnu/Linux is arbitrary changes to command line option semantics and syntax.

Changes like new vs old awk are a big deal in large systems environments. Sun made new awk nawk.  IBM which came to the Unix wars late, made old awk oawk and new awk awk. Gnu/Linux made gawk awk.

It's trivially easy to interpose a sym link or shell script in your path to erase such differences so that a common model works everywhere.  This was how I dealt w/ working on Sun, SGI, DEC, HP, Intergraph, Alliant, Intel, IBM and other systems simultaneously via an X terminal and a common home directory mounted via NFS.  My PATH started w/ ${HOME}/${ARCH}: and whenever I encountered a variation from common practice I made the appropriate adjustments.

Complex programs like compilers and linkers don't have a common set of features and options. So the foregoing will not work for them.  However, gcc works on Solaris, so basic compiling should not be an issue.

As for being able to compile and link a simple program.  All systems seem to have moved to not installing development tools be default.  There are several different package installation tools for Gnu/Linux so being generally able to administer and configure any Gnu/Linux box you encounter requires familiarity w/ several tool sets.

> I'm fully aware of the power of the command line and it is
> the command line that really makes me like Unix based OSes
> (including Linux). But making OI look well-polished with a
> fancy and easy to administer web-admin GUI that would
> encourage the average-Joe to use it as a home-NAS / virtual
> server is not a bad thing. That way OI would reach a higher
> penetration with a larger user-base and most importantly; it
> will get _free advertising_. To some extent the old adage "A
> good product markets itself" has some truth in it. But it
> must not only be good, it has to /look/ good so that even a
> less versed person will understand how good it is.

I've never seen a GUI that provided a clear model of the underlying semantics for any program, much less something as complex as system administration.  Perhaps that can be done, but I'm skeptical.

The current system administration model for Solaris is a mashup of old and new which can be maddening at times.  However, I've found this true of Gnu/Linux also.  Transition is slow, painful and necessary.

> 
> Another weak point of the OS is hardware support. While I
> understand that it might be a tremendously daunting
> undertaking to maintain hardware support for everything
> there is out there, on can strategically focus on key
> hardware components. My personal pet peeve with OI is the
> poor support for my AMD/ATI GPU, otherwise I think I think
> it is a good thing that there is a focus on HBA and NIC
> hardware, i.e. hardware that is essential for it to function
> as a file/webserver as was stated in a prior post. But the
> most important thing is that it is well documented so that a
> user who wants to start using OI will know *beforehand* what
> hardware to use.
> 

Hardware support was always an issue for Solaris x86.  It will be until such time as hardware vendors support device drivers for their own hardware.  nVidia does this but few others do.  So what is supported is whatever Oracle/Sun chose to use.  There is a chicken and egg problem.  Device manufacturers won't spend money on drivers unless there's a large enough market and the market won't grow unless there are drivers.

It is certainly true that OI is a bit rough around the edges.  This is why I still use Solaris 10 on my primary system and OI as my internet access host.  However, I find OI to be neither easier nor harder to deal w/ than MacOS X, Gnu/Linux in its multitude of variants or BSD in its incarnations.

If you setup a LAN w/ Fedora, CentOS, Ubuntu, Debian, OpenSuse, FreeBSD, OpenBSD & OpenIndiana systems I think you'll see that OI is neither easier nor harder.  It's just different as are all of the others.  With USB boot you only need 2-3 systems and bunch of USB disks.  I can promise you the experience will be very educational.

Have Fun!
Reg



More information about the OpenIndiana-discuss mailing list