Analysts familiar with Sun Microsystems Inc.’s Project Indiana say that as early as this week the company could reveal plans to revamp the OpenSolaris operating system by incorporating key pieces of Linux software.
Sun OpenSolaris To Become More ‘Linux-like’
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSAlert.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Solaris has a very good kernel foundation, yet Linux has a better userland. Of course this is a personal opinion, yet I guess this is shared by many people.
Thus, the only “missing” piece for Solaris World Domination would be a Linux (or BSD) driver layer.
“Linux has a better userland”
Linux doesn’t have a userland. Linux is a kernel which is why we can have distributions like Nexenta which is GNU/Solaris as opposed to GNU/Linux. Although I’m only interested in the GNU.
I realize that your answer is a very technical point (albeit a very good one), but the grandparent’s point still stands and is still (arguably) correct, even if his language isn’t absolutely perfect.
Even if it’s technically incorrect and can be confusing in some instances “Linux” is, to many the entire operating system; not just the kernel but the userland too.
He was very obviously referring to “Linux” as an OS and not as the kernel.
But he shouldn’t do that as he’s hurting all the involved parties (http://www.linuxmark.org/).
“””
like Nexenta which is GNU/Solaris as opposed to GNU/Linux. Although I’m only interested in the GNU.
“””
Then why don’t you guys finish up GNU/Hurd. Then you wouldn’t have to hijack threads about other OSes. Though I suspect you would continue to do so anyway.
I get your point, and totally agree with it. Problem is many of the GNU types complaining about linux don’t even know what Hurd is, let alone have learned from it’s failures. They just hear v3 == “anti-tivo/anti-DRM/anti-patents” and correlate it to openSolaris CDDL/v3, without really understanding the implications of embracing a kernel that will remain without any upstream support in v3-only permutations.
I applaud Sun’s consideration of GPL’ing openSolaris but at the same time I can’t help questioning their agenda, so I can’t help wondering about the motivation of that portion of the GNU crowd that is so willing to jump on the bandwagon to embrace a corporately-owned triple-license (triple including proprietary) kernel versus one that is pure GPL.
I agree. While it might help put a feather in the caps of GPL advocates, Sun did this for business reasons above all else. Novell(yes cyclops, I’m going there) embraced OSS because they had to. IBM embraced OSS because it was an alternative to Microsoft and an emerging market. None of these public companies get on board with OSS for “pure” reasons. OSS is a means to compete, primarily against Microsoft, on a more equal footing.
“I agree. While it might help put a feather in the caps of GPL advocates, Sun did this for business reasons above all else. Novell(yes cyclops, I’m going there) embraced OSS because they had to. IBM embraced OSS because it was an alternative to Microsoft and an emerging market. None of these public companies get on board with OSS for “pure” reasons. OSS is a means to compete, primarily against Microsoft, on a more equal footing.”
I’m only getting involved although, because I was named. I’m personally getting more jaded the longer I post here. You don’t see posts like you used to regarding GNU being a “hobby OS” in fact someone posted “toy OS” the other day, and I was surprised. Linux long ago killed the myth that anyone other than companies wrote Open source code in the kernel, this is getting to be the “norm” for userspace if its not already so.
I’m not sure why I was mentioned in regards the novell stuff. Patents is a topic I am weakest over. Which is the main point of interest with Novell. I just can’t see them being anything but anti-“advancement of technology”, I can’t even see the benefit for Microsoft, apart from it reinforcing their anti-capitalist agenda.
I entirely agree with you that Microsoft is using its Monopoly on the Desktop illegally to encroach on others profitable market, but they are above the Law. I find it bizzare on here in general, that people on the Microsoft Platform actually think this is a *good* thing and use all kinds of arguments to justify it.
What is interesting is Open-source is not just GNU its starting to appear on Microsoft’s own desktop, even in areas where Microsoft *bundle* and alternative. So I don’t think this is platform specific.
Companies are seeing *too* late, that open-source can be part of strategy against Monopolistic abuse. I find it funny that one of the largest reason against GNU is niche products from Adobe/Sony and other companies that do not have a presence on GNU when Microsoft is slowly eroding their Monopolistic products, by then its all to late lock-in has already set in.
I’m often surprised that open source does not have more *non-programmers* active. Open-sourcing any product must be terrifying, for some this means changing your entire business model. You can see why Apple is surviving as an alternative, but its simply hard you can can see this with the barrage of *new* licenses which aren’t successful or compatible with anything else. Many companies try unsuccessfully to *get around* Open sourcing and it does not work very well. Flash on GNU is a great example of this, and I with Gnash was ready now, but we are seeing binary applications from google/Ahead/Real which look weak next to native open-source solutions. I’m actually glad the trend that Coral started and Google continued is unsuccessful, which is to use Wine as a method of getting applications on GNU.
But the bottom line is your right. Companies often support GNU because they have been dragged kicking and screaming to the GNU table…kicked by Microsoft. This is possibly only a bad thing if they try and work around copyleft, becuase I don’t think it works. I suspect CDDL is a way of doing just that. Regardless of Linus’ insecurities he should learn to be a little fish in a big pond…and think big, and is becoming increasingly anti-business and a man I have less and less respect for. Having a *company* as opposed to an *individual* in charge of a project is a bad idea for anyone other than the company involved. The license does make Solaris more attractive it *will* be my next kernel, but I have no doubt that for success of solaris GPL3/CDDL may not be open enough for other companies or users of open-source software.
“””
Companies are seeing *too* late, that open-source can be part of strategy against Monopolistic abuse.
“””
I sometimes say to myself, in a half-joking way, that open source is where dying companies go to die.
That’s not a slight against FOSS, but a recognition that some companies seem to wait until their condition is terminal before opening up and seeking help.
It reminds me a bit of the adherents of the Church of Christian Science on my mother’s side of the family. Not “believing” in doctors, they pray, visit practitioners, and wait until their condition is terminal before breaking down and seeking medical care. After they pass on, it having been to late for modern medicine to do much for them, their friends collect together and nod, solemnly, seeing it as a confirmation that “those doctors can’t really do anything”.
That is… until it is *their* time.
Edited 2007-07-13 16:57
> IBM embraced OSS
Yeah right. Have a gooooood look at their product list. OSS is a very small percentage, and not that much greater than Microsoft.
Product list does not equal support in terms of dollars and directly paying developers.
Still, I would caution against making too many assumptions about IBM’s positions. They perceive OSS to be of benefit to them today, with their current strategy. And to the extent that we have a shared interest, we have an alliance. And that’s cool and beneficial to everyone. But it is important to remember that thier interest in OSS arises entirely from perceived *self-interest*, and could change due to circumstance, or change of management. And, of course, the alliance only extends to those areas in which we do have a shared interest. When it comes time to “reform” the patent system, don’t expect them to be in alignment with OSS authors on what new reforms to support.
i.e. pledges and whatnot are great. But I’d definitely get it in writing, and signed by parties with the appropriate authority.
You’re preaching to the choir! I agree with you completely. Basically what I said in my first post was what you’re saying. IBM, Novell, Sun, they’re on board with OSS in general right now because it’s in their best interest to be. That’s basically it as I see it. That doesn’t mean the community can’t benefit from those companies current positions, in a variety of ways, of course.
Yes.
I was just trying to say it in a different way. Plus, that’s my standard spiel about OSS’s alliances with those fickle, self-serving entities which we refer to as publicly traded corporations.
It’s kind of like having a pet snake. You can easily attribute all sorts of human personality traits to it. Even affection. And then one day it does something, like strangling and consuming your little sister, that reminds you, too late, that there is an alien, reptilian brain in there.
Some might not find that a terrible thing.
You’ll have a GNU/* when you have a distribution without Linux kernel, X.org/Xf86, Qt, KDE, Python, CUPS, Mono, samba, Gimp, OpenOffice, gstreamer, and on and on and on.
What nerve, actively trying to upstage everybody else.
Linux _the system_ is called simply “Linux” because *people* like it that way; the name is not an imposition. But GNU/* proponents are constantly bickering to *demand* pole position in the name, while happily forgetting attribution to the contributions of countless others.
Check “envy” and “pride” in your list of deadly sins. You might even be going to hell!
We have nothing to wory about as log as we have a user base with such a firm stance
@lobotomik
I gave you a civil response…but it didn’t deserve one. I refer to this quote.
“But GNU/* proponents are constantly bickering to *demand* pole position in the name.”
You may live in a world where FSF vs kernel developers, and you have to *pick* sides. I don’t I use software from *both*. I personally think its a disgrace that Linus encourages infighting between open-source/free software. When in reality their is the more critical 90%+ Market share; Patents etc that *he* should be concerned about.
His posts regarding Solaris/GPL3 after he had taken a lot of the bite out of the GPL3 license something everybody is saying GPL3 is a goof idea post Novell, and Linus clearly fears in your words a Linux without Linux, or I could say GNU/Solaris, but I wouldn’t I’d GNU. I say this because there isn’t a Linux developers vs FSF, because politically as we see there are shades of grey even amongst kernel developers and companies responding to a mythical community. Its not unusual to see code dual licensed BSD/GPL in the kernel for this reason. I suspect strongly that we will start seeing more code licensed GPL2/GPL3.
Please do not get me involved in any infighting between the FSF and Linus. Because I don’t care I don’t care which kernel I use as long as it supports *my* freedom.
@DittoBox he is referring to Linux the “OS”, but that the irony of what he has written. In post 2006 Vista where IM; Media Centers; Media Players; are part of the OS. Those very applications exist above the Solaris and Linux kernel. Do the difference he is talking about *is* the kernel. Its not 100% true.
@sherman14 I’m not sure which “guys” you are referring to I can only assume that you are referring to some sort of GNU thing so I will answer your point there is an *awful* lot of GNU that is *integral* to the functioning of OpenSolaris.
@Lobotomil I loved you post I esp like this “You’ll have a GNU/* when you have a distribution without Linux kernel, X.org/Xf86, Qt, KDE, Python, CUPS, Mono, samba, Gimp, OpenOffice, gstreamer, and on and on and on.”
The opposing argument is I only use Linux and OpenOffice from that list. The idea of naming the OS after the kernel is a outdated idea, but it doesn’t describe *my* OS. I run Gentoo predominantly becuase its a meta-distribution and this allows me to pick and choose my modules that make up *my* OS, and as we are seeing Linux/Solaris has become just another component, and an attractive one at that.
Linux is a genetic term for a *distribution* that runs on the Linux kernel. What is the genetic term for a distribution that can run on a verity of kernels? I use the term GNU only becuase it describes software thats predominately under the GNU license.
I’m personally bemused by Linus’s assaults on the FSF, simply becuase in every way *apart* from politically they sing off the same hymn sheet, and he used a political license.
I am happy for you to rename what I describe as GNU, my mythical meta distribution. If you come up with a better name I promise to use it. I hope you will as you care about so much about my soul.
Edited 2007-07-13 12:24
Building apps in Solaris is not always trivial (well, it isn’t *always* trivial in linux, but 99% of the time it is). Hopefully Indiana will largely alleviate this problem.
Why should project Indiana fix problems that aren’t with Solaris, but rather with people that make stupid assumptions about their software and make it non-portable?
Seriously. I would estiamate that over 90% of the problems I have encountered relating to software compilation and builds on Solaris have absolutely nothing to do with Solaris and everything to do with authors that write software for Linux and assume that every other OS works just like it. It’s retarded.
The same problems often happen on FreeBSD, etc. unless the authors have specifically accounted for them.
This is true.
I used Linux for a long time, but do to some odd circumstances, both of my computers (a Sun Blade 2000 and a Tadpole Sparcle laptop) are UltraSparcs with Solaris on them.
Compilation even with the GNU toolchain can be a serious pain in the butt when building stuff written to run on Linux. Of course, NetBSD’s pkgsrc which apparently supports Solaris does not work that great either.
Not to mention the guys that managed to write non-portable shell scripts, which I can fix, but the first time I bumped into it was an eye opener.
Probably what is occurring is that people write stuff using the only environment they have available, the GNU, but wind up relying on the GNU-specific options. Not evil, but just ignorant (in the completely non-judgmental version of the word.)
Linux is the kernel. Project Indiana is not touching the kernel of OpenSolaris. Hence, OpenSolaris will not be more Linux-like.
They are looking at the userland bits, which are not Linux. They’re looking at things like GNOME, KDE, applications, interfaces, frameworks. All the things that go above the kernel.
Basically, they’re trying to make Solaris look pretty, and (possibly, the article is sketchy on this) conform to Linux OS standards like the LSB.
IOW, they’re just going to start using more OSS apps/frameworks on Solaris.
Not really news. And has very little to do with Linux the kernel.
I still think that Indiana is about making OpenSolaris more Debian-like. This includes Debian-based package repositories and a distribution build system similar to those used by Linux distribution projects.
They brought in Ian Murdock, the perfect person to make this happen. Sun is increasingly close to Canonical, and if they were to support any of the packaging frameworks from the Linux world, they would choose Debian. Sun and Canonical would team up to bolster the Debian corner of the packaging triangle. They would put pressure on Red Hat and send Novell into panic mode (as if they know any other way).
Canonical needs ISVs to certify for their platform, yet most of them are content to stick with the big two. Sun gets ISV support, but they have problems integrating the latest and greatest from the free software community. If Indiana is what I think it is, both vendors can say goodbye to their respective problems.
I think we are all missing the fact “Linux-like” was probably dreamed up by some marketing guru who want to sell solaris to S/As and CTOs who are still trying to understand why Windows is “Bad”
Hell if they said They were going to keep the Solaris kernel and use NetBSD’s userland and apt I am sure it would make Solaris more “Linux-like” . The Linux bit here is a GeeWiz word of the day, not the Kenrel.
Ian make solaris more modular and, easy to manage with out a GUI; I will start using Solaris more . And Another thing why the F did sun decided to use Java for every blessed thing .
“why the F did sun decided to use Java for every blessed thing”
I too have wondered this. As my Sparcle is only a 650, I am a non-Java fan.
My best guess is that they do it as a marketing thing: “See, Java can do XYZ too!”.
Take the Sun SPOT. It is a microcontroller of somesort (I think). Sun wrote the *drivers* in Java. Just because they could, rather than it being the best language for the task at hand.
Horribly inaccurate.
Project Indiana isn’t about changing Solaris.
It’s about a new project.
Solaris won’t be changing for *years* from now since most of what Project Indiana is doing will likely require a major release binding.
Besides that, Ian doesn’t have the ability to decide what happens to Solaris. Only Sun engineers do, and he isn’t one of them…
Edited 2007-07-12 23:57 UTC
The way I saw it on the OpenSolaris project website, Indiana appears to be a OpenSolaris based distribution which attempts to make Solaris nicer to Linux and non-UNIX familiar people.
If it means attracting new contributors, new users and more attention, hopefully that’ll bring in more hardware and software companies willing to create not only server but workstation orientated software and hardware support as well.
One thing I would love to see is Crossover Office for Solaris, for instance – allow one to run Photoshop on *NIX would be an awesome thing to look forward to
This exists today…Mac OS X. It’s *NIX, and Photoshop runs on it natively.
“This exists today…Mac OS X. It’s *NIX, and Photoshop runs on it natively.”
Mac OS X had little or nothing to do with its FreeBSD lineage. The reality is Mac what makes a MAC popular isn’t anything to do with *NIX it *is* about Carbon and Cocoa APIs or the Quartz Compositor and Aqua user interface.
I’m tired of this lie.
Huh?
GP said “I want to run Photoshop on Unix”
Parent said “Run it on Mac OS X: you get Photoshop natively, and it’s Unix”.
You’re saying “It’s not FreeBSD”.
Nobody in the thread claimed it was. Mac OS X isn’t FreeBSD (though it does share big chunks of code with it). Who cares? SVID-compliant, trademark-licensed Unix. With a TM, if you like. And it runs Photoshop natively. Apache + MySQL + Postgres + Unix tools + Photoshop, Lightroom, Illustrator + Nice GUI, etc, etc.
It’s not to everybody’s tastes, but it is exactly what the GP asked for!
I’m not sure why on earth you had to make an argument out of it.
And you know, ever thought that maybe I don’t want to get rid of my WHOLE computer/laptop simply so I can run Photoshop? Why the heck would I want to pay NZ$2499 for a laptop (Macbook) which lower specifications than my current laptop which I bought at the same price.
Ohaye don’t get me wrong: I’m not advocating Mac OS X specifically, my beef was purely with the “I’m tired of the FreeBSD heritage lie” post.
I too have to appologise for replying to the wrong post – I was directly my original reply to the person whom you replied to.
I too, like you, get angry about this link people make between MacOS X and FreeBSD – just as it pisses me off when people try to equate Linux with UNIX (or vice versa).
If people knew the history of Mach, I don’t think they would make such stupid claims – to say MacOS X is based on FreeBSD would be like saying OSF and Tru64 are based on FreeBSD by virtue of them being based on top of Mach.
There is a strong link between FreeBSD and Mac OS X as there is between Linux and Unix. I am not interesting in arguing how much.
http://developer.apple.com/technotes/tn2002/tn2071.html
Its not my link, but its quite fun has this fun quote.
I would argue that that the Mac OS * and *BSD are two different beasts, but they do share a lot of code. I am simply not qualified to say how much.
Edited 2007-07-14 14:33
“There is a strong link between FreeBSD and Mac OS X as there is between Linux and Unix. I am not interesting in arguing how much.”
For origins of Apple’s Mac OS X and Darwin, I recommend having a look at the BSD family tree, /usr/share/misc/bsd-family-tree on any FreeBSD installation, or, if not at hand, see:
http://cvsweb.netbsd.org/bsdweb.cgi/~checkout~/src/share/misc/bsd-f…
Or consider having an interesting read in: Goodheart, Berny & Cox, James: The Magic Garden Explained. Prentice Hall of Australia. Sydney etc. 1994. S. 3ff.
@Doc Pain
Thank you
Double post!?
Edited 2007-07-14 19:44
Ian doesn’t have the ability to decide what happens to Solaris. Only Sun engineers do, and he isn’t one of them…
You’re partially correct. Ian Murdock’s title is Chief Operating Platforms Officer. So he technically isn’t an “engineer,” but he is able to decide what happens to Solaris, at least to a certain extent.
Wrong; if you read things posted by Sun engineers, you will see that upper management doesn’t get to control the technical design of their products.
a GNU/Solaris system, the gnu userland is just so very much more pleasant to use, and that is after all what one uses to interact.
about kernel? well i couldnt really care much, though if solaris goes gplv3, that is a major reason for me to want it, also i believe torvalds has been making some terrible decisions for linux, which is certainly not helping to make me want linux.
So if debian gnu/solaris or gentoo gnu/solaris comes, i think linux will have a pretty hard time justifying me installating it instead…
I am curious to know which decisions of Linus’ you think are poor? Technical ones or his disagreement with the FSF? I am not trying to provoke anger or outrage. I sincerely am curious, because I have not heard many complaints regarding technical decisions.
for one the decision to choose CFS over SD.
You are kidding me. Can you point out anything technical wrong with CFS?
CFS is inferior to SD.
ingo continually fails to fix the smoothness problem.
Isn’t this what Nexenta is all about?
I hope we aren’t seeing a pointless duplication of effort…
OpenSolaris is what will eventually become a future, fully supported Sun Solaris release. That’s something Nexenta will never be able to claim
Nexenta seems to be adding a Debian variant of the GNU userland to Solaris, while I’m sure this project will be much more of a hybrid of Solaris and GNU.
Well a GNU/Solaris combination is appealing to those who believe the Linux kernel is poor and the GNU userland is decent.
I think the Linux kernel is the best of all Unix and Unix-like Monolithic kernels because of it’s constant overhauling and feature additions that suit modern requirements. Lately it has been the new wireless network stack and process scheduler.
Solaris is nice but it has nothing to gain by being another Linux.
Solaris works great the way it is right now. It is an great server software and does not need the linux stuff.
Please do not make a linux of it.
Insinuate that Linux-related has anything, anything of all, of value to offer Solaris.
Sheesh.
OK, I’ll bite. What does Linux have to offer that would make Solaris easier to use? There are a number of areas that Sun can improve on, but I feel that just making Solaris (or other variants) “just like Linux” at appease those who only know Linux as the wrong approach.
Where I get pissed off is people bithcing about how hard it is to use Solaris simply because they haven’t taken the time to learn anything about it. I read a blog post about a Linux user who got a T2000 on a Sun “try and buy” and because he couldn’t find top he loaded Ubuntu! Obviously he has never heard of prstat (which with the -a option provides similar output to top).
One of the discussion items on OpenSolaris is about replacing the root shell with bash. My question is why? If you are administering your system(s) correctly and use least privilege concepts then you should spend minimal amounts of time as root. If you are spending that much time as root that you need the features of bash or any other shell with advanced automation features, you shouldn’t call yourself a UNIX admin.
I think the “usability gap” is nothing more than (1) an issue of user education and (2) people who simply don’t want to change their way of doing administration. And do I think Sun should “cave in” and change Solaris simply because a bunch of Linux users don’t want to change, NO!
OK, I’ll bite. What does Linux have to offer that would make Solaris easier to use? There are a number of areas that Sun can improve on, but I feel that just making Solaris (or other variants) “just like Linux” at appease those who only know Linux as the wrong approach.
Where I get pissed off is people bithcing about how hard it is to use Solaris simply because they haven’t taken the time to learn anything about it. I read a blog post about a Linux user who got a T2000 on a Sun “try and buy” and because he couldn’t find top he loaded Ubuntu! Obviously he has never heard of prstat (which with the -a option provides similar output to top).
One of the discussion items on OpenSolaris is about replacing the root shell with bash. My question is why? If you are administering your system(s) correctly and use least privilege concepts then you should spend minimal amounts of time as root. If you are spending that much time as root that you need the features of bash or any other shell with advanced automation features, you shouldn’t call yourself a UNIX admin.
I think the “usability gap” is nothing more than (1) an issue of user education and (2) people who simply don’t want to change their way of doing administration. And do I think Sun should “cave in” and change Solaris simply because a bunch of Linux users don’t want to change, NO!
Well, I didn’t say, or even imply, that Linux had anything to offer Solaris, just that the mere implication itself was enough to send Solaris fans into a litanies and diatribes. And sure enough… And this one was rank with UNIX snobbery of the highest order.
So here’s a question, if you had your druthers, what, if anything, would you do to Solaris to expand its market? Or would you even want the market expanded?
That figures. If this “diatribe” was posted by a Linux user it would be modded up to 5. You can call it whatever you want, the bottom line is what I said is true.
While I think Sun should do some things to expand their markets, changing Solaris to make a number of Linux users who think Solaris is “hard to use” isn’t one of them. If AIX and HP-UX ran on x86 hardware like Solaris, I would expect to see the same kind of comments about them not being like Linux as well. At some point one actually has to learn how to use the OS and not just mindlessly complain about how its easier to do something in Linux.
The process should be one where there is an actual benefit for the majority of users and not just a small number of users who might or might not use it or benefit from it.
“””
You can call it whatever you want, the bottom line is what I said is true.
“””
In what context? No one has bothered to even describe what this machine that we are talking about is going to be used for. Is my grandmother going to be using it to read email? Is it in a high school computer lab? Is it my own home desktop system… or server?
Robert, as much as I respect you, I have to kind of agree with the OP’s assertion that some Solaris guys have a knee-jerk reaction about these things.
My position is that for general purposes, both the Linux world and the Solaris world could benefit from bits on the other side of the fence. And it really all depends upon the context.
Sun would like to see Solaris competing in arenas outside its traditional usage. And so would I. Because, at heart, I’m an advocate of the Unix way of doing things. Linux simply suits my purposes best today.
But I have huge respect for Sun (and Scott!) for holding their ground, and for believing in… for standing behind… Unix, while HP and even Silicon Graphics were pushing Windows.
And, Robert, I sincerely hope that arrogance, from either side, does not drive another wedge between members of our Unix world. Because we have had enough of wedges. They’ve cost us, dearly. We can’t afford any more.
And I sincerely hope that we do not have to endure even one more of them. Especially now that Sun is walking the walk. And quite stylishly, I should say.
To my fellow Linux fans? Let’s not go out of our way to start a fight.
Edited 2007-07-13 21:01
I have read more than my share of posts here and other places talking about how various people cannot use Solaris for one reason or another, and it is arrogant to point that out as a possible reason why so many Linux users want Sun to change Solaris is simply to suit them?
I am sure that similar comments can be found in any number of forums (including this one) where Windows users would like to see a more user friendly face put on Linux, and the response by certain members of the community would be no different.
So am I really all that arrogant, or is it because I am willing to voice my opinion which might piss a few people off? And that is different from any number of people who post here how?
Robert,
I was afraid that my post might have come off as calling you arrogant. That’s my mistake. Not really what I intended.
Windows certainly *is* perceived as being more user friendly than Linux. And in some ways it very much is.
In others, Linux actually does better in an objective sense. But people like what they are familiar with, and for most folks, Windows is it. (Not to mention the unnatural degree of ISV support.)
So please focus on the part of my post which is about drawing upon each other’s strengths rather than fighting over differences. Because I *really* *really* *really* do think that is important.
And as big a Linux fan as I am… believe me… I am solidly in favor of the Unix way more than anything. And if I felt that Solaris might be a better fit for my campaign, I might be inclined to make the jump sooner than I should. Or… I might find, in restrospect, that I had clung to Linux, out of familiarity, longer than I should have.
I really don’t know.
But I *do* know that petty conflicts are detrimental to us all.
Thanks, and +1 for you. While I agree that petty squabbles get us nowhere, but unfortuantely I see the “usability gap” as potential for Sun to waste a lot of time trying to make some people happy who might not ever be happy.
Usability of anything is based on education and experience, and what I see and hear is people who have very little experience with Solaris making a lot of noise and instead of actually learning how to use Solaris, they want Sun to change Solaris so it fits their experience level.
The link provided by jwwf exemplifies part of my point about usability from the perspective of a user who is determined not to learn anything about the OS. It is not that he could not have downloaded top from any number of sources including SunFreeware and Solaris4u or compiled it himself if he could have figured out how to add /usr/sfw/bin to his PATH. Funny, I have no problem using gcc on Solaris 10 or Solaris Express with seconds of completing an installation. Creating a PATH is part of UNIX 101 and if somebody cannot do that they quite frankly should not be using UNIX or Linux.
Another is the recent article linked here about the “Solaris Installation Experience” by Michael Larabel of phoronix.com. When an article starts out with “There’s a problem with Solaris and Sun knows it. The installation experience of Solaris (along with other areas) could be greatly improved.” and then compares an installation of Solaris Express to Fedora 7 that ends up pointing out nothing more than semantical differences between the two operating systems does nothing for anyone.
A more insightful article would of been to compare a Solaris 10 to Solaris Express installation to see what Sun has changed to make it easier for neophyte users. From the viewpoint of an experienced Solaris administrator much has changed with Solaris Express. A Solaris 10 installation gives you six choices including Custom JumpStart and two text modes, where the Solaris Express installation gives you two. Solaris 10 does give you the ability to manage packages and dependencies during the installation while Solaris Express does not. Considering the learning curve of understanding six install metaclusters for Solaris 10 and the certain dependency Hell associated with that level of complexity, it only makes sense for Sun to minimize choices to what is actually necessary to get a working system with the least effort. This was not pointed out in the article because Michael didn’t evaluate Solaris 10 against Solaris Express.
Several OSAlert readers have complained about the graphical installer as well. Here’s a tip, right click on the background and you get a menu that you can use to launch a terminal session. Among other things you can use that terminal window to reset the root (or any other user’s password).
For those who want full control of the installation process I would use Solaris 10. For those who want to work with bleeding edge features or want to work with OpenSolaris use Solaris Express.
My concern is that Sun is going to waste a lot of time trying to make people who can’t take the time to actually learn Solaris happy by adding a lot of Linux features rather than make better documentation available. I would rather see Sun concentrate on improving specific features of Solaris that appeal to a large number of users then try to win “mindshare” by attempting to cater to a handful of users who basically have no intention to spend a dime on any Sun product or service. I learned a long time ago that the customer isn’t always right.
My comments are the result of years of experience working with Solaris from 2.5.1 to 11, including 4 years of Beta testing. At some point the users themselves have to take responsibility for their actions and I don’t see the point in Sun bending over backwards to please people who do not want to learn, especially when the information is readily available.
I would suggest that the Solaris-related communities simply do what makes sense for them, while keeping in mind that needless and arbitrary differences put up walls that reduce the overall productivity of the universe in which they exist.
As a real world example, look at the standardization of the world on the English language. Now, I would *never* claim that English, my own native language, is superior to French, Dutch, German, Italian, or Russian. (I’d take Esperanto any day, but no one seems to care about it.)
But imagine an internet which had not standardized upon a language, with everyone speaking the language of their own country. And if a German-born programmer and an Itialian-born programmer wanted to converse, one or the other of them would have to learn the other’s language. And if they then needed to collaborate with a fellow from France… well… things would get complicated very quickly. And the fruits of collaboration would be catastropically curtailed.
So, if Solaris’ implementation of ‘tar’ is actually superior to GNU tar, then by all means, keep it. But if it is just *different*, why not move toward a common standard?
Sun really does not have to waste effort trying to please the Linux world. They simply need to look to what would benefit them, and the rest kind of falls out of that, naturally.
Edited 2007-07-14 18:00
“And if a German-born programmer and an Itialian-born programmer wanted to converse, one or the other of them would have to learn the other’s language.”
Just imagine the fun an English speaking programmer would have with today’s whizkids in Germany and their excellent Newspeak English… I you driwing to fast you, become a letter from uss and paying! int bratwurst;
“And if they then needed to collaborate with a fellow from France… well… things would get complicated very quickly.”
Charsets are fun. While you can even emulate Russian letters using the standard (!) ASCII characters, German and Frenchmen have nice combinations of vocals with ‘, `, ^ and : (rotated 90°) on top. int müsli, eisb"ar_knut, d"oner, lochfrass; /* Will it be displayed correctly? */
“So, if Solaris’ implementation of ‘tar’ is actually superior to GNU tar, then by all means, keep it. But if it is just *different*, why not move toward a common standard?”
I agree with this statement. Exempli gratia, FreeBSD actually does not use GNU tar, but BSD tar, which offers compatibility regarding command line parameterization and functionality to GNU tar of course. It’s just a different implementation on the long way to get the BSD userland into BSDL status, but to the user, it does not make a difference, it’s just like the inner parts of an API changed, but the interface keeps the same.
A short comment on another point made in the article: Those of us who use Solaris on a daily basis will know for sure: Boot up time does not matter because you don’t reboot your Solaris installation day by day. Usually, a Solaris systems runs days, weeks, even years without reboot. So I think someone who’s half the way a professional can stand the startup (which does, compared to BSD or Linux, take some time).
“OK, I’ll bite. What does Linux have to offer that would make Solaris easier to use? There are a number of areas that Sun can improve on, but I feel that just making Solaris (or other variants) “just like Linux” at appease those who only know Linux as the wrong approach.”
Maybe this is the same development that makes Linux “just like ‘Windows'”? Because users are assumed to “want it that way”? Abandoning features to make users of other OSes feel familiar? Hmmm…
Are there really demands explained to the OpenSolaris team so they are going to make these changes? Making OSol more Linux-like would include massive changes, I think…
“Where I get pissed off is people bithcing about how hard it is to use Solaris simply because they haven’t taken the time to learn anything about it.”
User: Learning is dull. Everything has to work out of the box under any circumstances and by itself. Delegating problems to others is smart.
“I read a blog post about a Linux user who got a T2000 on a Sun “try and buy” and because he couldn’t find top he loaded Ubuntu! Obviously he has never heard of prstat (which with the -a option provides similar output to top).”
This isn’t for real, is it? I heared of such kind of behaviour in MICROS~1 land, but in Linux world? Hard to imagine. Try this one: “I just installed this ‘Vista’ and my pictures are gone, I don’t find nothing! I’ll install my good ‘XP’ anyway.” (The term “pictures” refers to icons and dialogs and is a famous term among Germany’s PC users suffering from functional dyslexia.)
If I wanted to insult someone, I’d say: “Puh, Linux users are getting more and more stupid as their OSes advance…”, but I’m a polite guy and don’t say this honestly.
“One of the discussion items on OpenSolaris is about replacing the root shell with bash. My question is why?”
The root shell may always be sh (the bourne shell) because it isn’t often used in interactive mode. Instead, it’s used in batch processing mode (e. g. for startup scripting) where no special bash features are used. Today, most #!/bin/bash declared scripts could be run with sh without problems.
Even FreeBSD uses sh as the default maintenance mode shell, and the system scripts are declared to be #!/bin/sh. The Bourne Again Shell can be installed as additional software which will require #!/usr/local/bin/bash for scripting. The standard dialog shell C Shell (csh) of today does most interactive stuff equally or better (in regards of completition behaviour) than bash, even sometimes a bit tweaking is needed, such as set promptchars = “%#”, set prompt = “%n@%m:%~%# ” and set autolist.
“If you are administering your system(s) correctly and use least privilege concepts then you should spend minimal amounts of time as root.”
This is correct. The dialog with sh is usually needed in maintenance mode (single user mode) only.
“If you are spending that much time as root that you need the features of bash or any other shell with advanced automation features, you shouldn’t call yourself a UNIX admin.”
I couldn’t have found a nicer example. But today’s Linux users aren’t UNIX administrators.
“I think the “usability gap” is nothing more than (1) an issue of user education and (2) people who simply don’t want to change their way of doing administration.”
As I mentioned before, users don’t want to change. Now the difference between user and administrator enters the scene. In UNIX world, the difference is obvious and well intended. Since MICROS~1 products have taken over the home computer market, the difference seems to be disappearing, but in fact, users don’t do their administrative tasks, assuming the OS or someone else doing it for them. You know the result well: Spam, viruses, worms, trojans, illegal file sharing, repairing amd maintenance time++.
“And do I think Sun should “cave in” and change Solaris simply because a bunch of Linux users don’t want to change, NO!”
I think there are many tools insode Solaris that make administration easy. There are even GUI applications or remote administration tools that can be used. Who would use the silly keyboard and a strange 80×24 monochrome shell where he needs to write things instead of clicking on pretty pictures?
To get on topic again: Usability is a thing that is very very important to users, but not to administrators who have learned to use the tools that they have (and to create tools they don’t have). If Solaris will include bash in the future, along with other Linux tools, it will be okay, as long as the traditional ways of doing things are still available. They are the easier way to most administrators, because they have got the knowledge and the experience to use their power. It is their responsibility to keep the system running, and exactly this requires this knowledge and experience.
You can hand over tons of power to a user, but it he does not know how to use this power, he’ll be lost anyway, no matter how shiny the GUI is and how much sqeaking buttons and dancing elephants it has.
Solaris has had bash as a shell for years (Solaris 9 4/04):
# cat /etc/shells
/bin/sh
/bin/csh
/usr/bin/csh
/bin/ksh
/bin/tcsh
/usr/bin/tcsh
/bin/bash
/usr/local/bin/tcsh
/sbin/sh
The blog I was referring to is real, but I lost the link. I could not believe it when I read it myself. I wouldn’t joke about something like that.
While I agree with you about basic use, some of the more vocal uses here are doing things that an administrator or advanced user would do, such as compile software. And knowledge of Solaris from a developer’s perspective is important to avoid gotchas like “unable to create execuatable files” errors because the user has put /usr/ucb in their PATH, or don’t know that in a Full Distribution install of Solaris 10 or the default install of Solaris Express Developers Edition gcc is included (/usr/sfw/bin/gcc).
Nice post, +1 for you.
The blog I was referring to is real, but I lost the link. I could not believe it when I read it myself. I wouldn’t joke about something like that.
http://blogs.smugmug.com/don/2006/08/15/sun-fire-coolthreads-t1000-…
I think this is what you are looking for. I found it upsetting that some LAMP wizard did not know how to type ‘find / -name gcc’
That’s it, thanks! Isn’t that something?! I know from working with a “try and buy” T2000 that a Full Distribution installation of Solaris 10 is done, all he had to do is add /usr/sfw/bin to his PATH.
I tend to agree with Linus Torvalds: Apart from zfs there isn’t really anything there for a long time Linux user. Certainly, the kernel is good, but I don’t regard it as highly superior compared to the Linux kernel.
– The Solaris userland is, well, not from this world anymore …
– The default configurations are abysmal – to say it in a very polite way (it certainly would be very easy to change that)
– JDS is ok, but compared to GNOME in SUSE or Ubuntu lacks various important features
– The system as such (well, the latest OSOL builds anyway) is dead slow compared to Debian. And I am using a SUN x86 workstation.
– Automatic comfiguration of the nVidia driver is nice, but senseless of it fails. On my machine – as I said: a x86 SUN workstation – it fails utterly. Debian has no problem on that very same machine.
Let me put it like this: In five years from now I’ll return to Solaris. I expect it to be mature and grown up. At the present moment there is nothing – apart from zfs (and zfs is not incorporated into the installer) – which makes me want to try OSOL on a regular basis.
Re-visit it after one or two years – at the current speed (based on my judgement from reading the OpenSolaris lists), OpenSolaris will be a very interesting OS to play with by then.
Edited 2007-07-13 16:27
If Solaris were licensed under GPLv3, can they use Linux code, or would they only be able t use the parts of the kernel that are licences v2 or later, which is evidently not all of the kernel.
It seems that since GPLv2 is copyleft, it is expressly incompatible with GPLv2.
Funny one that.