Slashdot Mirror
Sun Java Desktop 2 Review
Anon. writes "Linux.com is carrying a pretty damning review of Sun Java Desktop System version 2. JDS seemed to have issues with almost each and every machine the author tested it on, support was quite bad - and to top it all, the software comes with a seven page license document. Something seems to be terribly wrong somewhere - otherwise why would Sun decide to ship JDS with kernel 2.4.19 at this stage?" (Slashdot and Linux.com are both part of OSDN.)
meh..i would have thought that they are one of the few people who stand to lose out from linux becoming more usable
I have discovered a truly remarkable sig which this post is too small to contain.
Trouble with hard disks, especially SATA but also regular ATA, seems to be a common problem this guy is having. That should not be a problem with any modern Linux distro, and why Java Desktop manages to screw it up I suppose we'll see.
I'm waiting for the next version.
The Cheese Stands Alone.
I guess Sun deserves what it gets, but I think the reviewer was a bit irresponsible. Perhaps he had a deadline and couldn't wait around for replacement media (assuming that you still couldn't rule out defective media) or for Sun support to resolve the issue. I think however that it would have been a much more useful review if the reader found out exactly why the reviewer couldn't get it installed on all but one machine and couldn't get it to run on the machine on which it did install.
I'm left wondering if it wasn't in fact defective media, and just how bad Sun's support is: meaning, what does it take to get a problem resolved.
quiquid id est, timeo puellas et oscula dantes.
Sun's pretty much a dead company walking.
Their hardware is more expensive, and slower.
Their OS is less feature rich, but has more bugs, and doesn't perform as well in most cases as Linux.
Look around, everyone who possibly can is getting off of Suns and onto Linux x86. The major things holding most of Suns customers back in this regard are proprietary software support, and that's improving all the time.
And as to Java... I'm not sure exactly how they intend to make money there... IBM does the Java services market SOOOO much better than Sun does.
In the past, I've had lots of problems with ATA hard drives, as well as controllers, in Solaris x86. YMMV, but I've only successfully installed it on two controller/drive combos. The five or six that failed were nowhere near top-of-the-line, but they weren't generic cheapo product either. I dunno. Long-standing problem with Sun x86 products, though.
The Linux distro scene is in a rather bad state ATM:
:-(.
1) Debian sarge release was pushed further on - and you have to go via knoppix to install Debian on a modern SATA machine, leaving the system in a messy state. Obviously the Debian (non-)release is a standing joke, but Sarge will be so late, it's not even funny anymore
2) FC2 was released, and it has several showstopper bugs (it keeps on crashing for me, it eats partition tables for dinner, keyboard layouts don't work, etc, etc). I'm sticking with FC1. FC1, OTOH, seriously rocks, once you beef it up with KDE 3.2 and kernel 2.6. FC1 is the best Linux I've ever used, and I was hoping it wouldn't stay that way after FC2
3) Suse is still non-free-beer. Come on Novell, letting hobbyists dabble with it at home isn't going to hurt anyone.
So what's left, then? Mandrake, Gentoo? Warez version of RHEL? WBEL?
And on the topic of JDS: they are always thrashed in reviews, but the media keeps hyping how "integrated" the system is, and finally Linux is of commercial quality. Go figure.
Save your wrists today - switch to Dvorak
With all this abysmal hardware support and horrid tech support, I often recall that Sun has made many a conficting statement in the past, and also that Solaris, Sun's glowing flagship, is still Top Dog as far as this 7-page-license company is concerned.
I would believe that running JDS on linux is just a prototype, to generate hype especially among the linux crowd.
Sun seems to be going out of its way to implement this on Solaris...perhaps the final incarnation of JDS will be on Solaris itself, without any Linux or GNU code at all, and completely proprietary. It seems that they will say "Linux just isn't up to par, but if you upgrade to the $599 Solaris JDS, all your hardware will work."
Or so it seems.
Defenestrate Windows...
- Not science folks like me, given the lack of (f2c or g77) fortran, which means no Octave for my analysis and no R for statistics.
- Not home users, given the non-inclusion of e.g. a working movie viewer. (Their java media player was completely busted -- it showed a few frames and then died.)
- Not cutting-edge linuxers, given the use of
the 2.4 kernel.
- Not the home market, given the use of
soffice (aka openoffice) which still won't handle complex msoffice documents well, and given the use of a stumbling movie viewer.
- Not future java-app users, since the java
apps included (movie viewer, text editor) are ugly and slow.
Having noted the above in my own tests, I switched back to Fedora for my home box [my work boxes are osx and solaris]. Using fedora [core 2] gives me (a) a newer kernel, (b) newer versions of software such as openoffice and mozilla and (c) easier updates.The real advantage may be in work-groups that have loads of existing Suns as well as linux boxes; there is benefit in having a similar GUI and similar software on each. This reveals the answer to the question of my subjectline, I argue.
I was there when the reviewer did the install, and I believe everything I write.
I tried different hard drives on different computers. I tried four different configurations, three of which were distinctly different. Sun could have easily updated the kernel in JDS2 but they chose not to.
It's not my fault they shipped a prehistoric kernel. Did you expect me to write a puff piece on an operating environment that doesn't work on all of my test equipment?
-Jem
We call this hugetlb, or did. The name keeps changeing. All this really means is that you support shared memory locked into RAM, hopefully with some sort of large-page notion.
a performance counter interface
We call this oprofile. We also have the user-space kcachegrind and the perfmon patch, so you get some choice on Linux.
hot-swappable CPU support
We have this too. It works great on IBM's zSeries mainframe. Oh, PC hardware? Solaris can't do that either because the hardware will die if you go ripping out a CPU.
solid device driver interface
Nice non-factual FUD there... you work for Sun? Linux has a sane, clean, simple, and high-performance driver interface. This is because the kernel developers don't contort the design to be more tolerant of binary drivers.
And the future is multiprocessors...
Your "beyond 100" is nothing to a 512-way SGI Altix running Linux. There are 1024-way systems under development. Sun can't touch this.
So basically, you would like us to believe that because Sun is (with justification) not very popular with the Linux community these days, that a reviewer is going to fabricate a complete lie and put his name it? Not only a complete lie, but one that is easily disprovable because he gave his exact hardware configuration and said "It will not install on this." Anyone with the same hardware (and a lot of people have things like that) can try it and see. If they try it and he's shown to be lying, not only his own journalistic integrity but that of linux.com, is shot.
It's not that reporters don't sometimes stretch things, or even flat-out lie. But they do it in ways that they are not likely to be caught. What you're describing is like a reporter going on the news and saying "The Washington Monument was painted pink today by the National Park Service." That statement is easily verifiable as such by many people, and for that reason no reporter would make that claim, even if he had a private agenda that wanted the Washington Monument to be pink as a means of embarrassing the government or the National Park Service.
If you read the entire review, you would have also noticed that the author saved his non-technical criticism (licensing, etc.) to the end, and included with it some advice to Sun about how to improve those problems. Hardly the actions of a person with an anti-Sun agenda.
You seem to be asking us to believe that because Sun has little political capital in the Linux community these days, that any Linux-oriented publication that does a review of a Sun product and has negative things to say about it must be lying, that these negative things couldn't possibly be true or accurate. I wonder, what is your agenda for making such a ridiculous claim? Do you work for Sun? Own their stock? Or perhaps you're really a Windows booster out for a Saturday afternoon troll?
Your claim is foolish. The people who modded you Insightful, doubly so. A wild and totally unsupported conspiracy theory claim is hardly the stuff from which Insightful posts are made.
yeah...
I have not tried the JDE, but i wasn't able to get Suse 8.1 and even Suse 9.0 installed as delivered on my machine due to SATA issues. Pretty much same error - installer couldn't find a place where to drop the OS.
My suspicion is that it may be the same issue.
SATA only got seriously addressed at the end of 2.5 kernel tree, as i recall Redhat had its own version of kernel 2.4 that incidentally supported serial ATA, but Suse was behind on this front. Of course, for Sun not to include such support is just a shame, since more and more users switch to Serial ATA storage.
Just my 2c
Using subtle clues and hints in his first-person narrative to imply emotion and intention, Jem Matzan's critically acclaimed writing style is truly unique among fiction authors. Jem's extraordinary characters and distinct dialogue decorate his fantasy universe while coaxing readers' imaginations into providing the specifics.
Also a professional actor and model, Jem spends much of his time performing in such productions as television commercials, stage plays, and interactive variety shows.
Biography provided by the author, October 2002
from here : http://www.scifan.com/writers/mm/MatzanJem.asp
In our experience learning OOP using Java is very difficult. All right, the bulk of students aren't that bright, but I think they would do a lot better if they learnt Python or SmallTalk using Squeak first. We kill a lot of interest in programming by using Java as the first programming language.
SATA drives are a problem with this kernel, but I have no idea why this reviewer had so much trouble with ATA drives. I've seen installfests where dozens of users installed JDS on _laptops_, all were successful. I've never seen it fail on a non SATA desktop.
SPARC contexts run out. Then you are faced with more registers than x86 or x86-64 has to deal with. The register stack engine isn't a bright idea; see the intro of the Alpha architecture manual where the DEC engineers tear into it. You're saving more, or less, context than you need to save. This is wasteful. When you run out of on-CPU resources, you take an expensive trap. What's more, recent chips from AMD have a hidden context mechanism that works entirely without OS help.
The features may or may not be cool and all, but it's the end result that matters.
You'd think this following statement he made would have been a clue that there was probably something wrong with the way he partitioned the 80 GB drive he claimed should have been able to handle the install:
The guy says the partition configuration he used for the JDS didn't work for SUSE 9.1 either, and had to repartition it to have enough space to complete the install! It appears to me that the only difference between this and the problem he had with JDS was that it sounded like JDS gave the warning a bit earlier in the install.
I believe that a JDS install should alleviate the user as much as possible complications from partitioning. Yet, we don't know if he had used "advanced" options that permitted him to create the limitted partition sizes, or a default install. In fact, we know nothing about the options he was given and the options he chose for partitioning.
Open Standards Portal
Debian/Testing is quite current. Debian/Unstable is as bleedin'-edge as they come.
Gentoo's nice too, but don't pick on Debian because Stable is old. Stable's not meant to be used by people who want the latest and greatest, it's for people who want a system that they can set up without things changing nightly.
The statement was the disc was partitioned without problem then there was an error message saying there was lack of sufficient space. Could one of these partitions been too small for the installation?
At least one of the responses claimed they had no problem with what seemed to be recent hardware.
I am no fan of Sun, but I will await other confirming assessments prior to fully believing their product is a complete disaster.
Probably two reasons: IBM has been generous, including GPLing some code (JFS, maybe more) and putting many people on kernel work (see the output:And we definitly don't like SCO even though they are a little guy. My guess a major part of the animosity shown toward Sun is that they won't GPL Java (see RMS on that issue).
But who knows?
My personal opinion is that the first thing Linux will kill is propritary Unix (Mac OS X is helping here, too). Then the real battle can begin.
Fellowship 9/11
My comments on their Live CD - here and here.
> It's not my fault they shipped a prehistoric kernel.
And since you knew that, why didn't you try JDS on an older PC? After all, JDS is SuSE -- you know that it will install just fine on 90+ percent of the office PCs out there.
And if you _had_ installed it on an older PC, then you could have actually _reviewed_ JDS (which would have served your readers), and you could have explained about the older kernel (also serving your readers), and then you could have complained about that specific point, instead of trashing the whole thing.
But what you did was dishonest, and did'nt serve anybody (except Microsoft, of course).
I think you purposely chose hardware that was too new to be supported by the Linux kernel included in Sun's JDS.
I think you're just another damned FUD writer.
I agree with choice, but I agree with gparent, we have enough redudancy. I might agree with you if every problem inherent with linux were properly solved, but they're not. At this point, we have far too much wheel reinvention, and that's not generally a good thing.