Impressions of Karmic Koala

I have been using 32-bit Karmic Koala KUbuntu for a while now. Given that it is supposed to go to actual release some time in October, I thought I should share some impressions.

It’s not just the color schema which has changed. KUbuntu actually looks and feels like Canonical might have put more than their customary $20 into the development effort. From what I’ve read with the bug reports though, the 64-bit edition got the same $1.98 in development funds it always gets. You still cannot install the 32-bit Lotus Symphony Debian package on the 64-bit edition of Ubuntu. (OpenSuSE 11.x fixed this problem with the RPM, but the Symphony developers chose a rendering engine which had a lot of bugs on the SuSE platform.)

There’s a lot of really neat stuff which has been added. The move to 00 3.1 was very welcome. The talk of dropping sun-java-6 is not. Many things actually work better. The newer version of SciTE was quite welcome as was the current versions of Tea and Leafpad.

Some things are just as busted as they have always been. Both Konsole and eTerm are still claiming to be VT terminals when neither really are. Until you can map the NumLock key as the Gold Key when in application mode and not have the emulator continue using it as a toggle between application and keypad mode, you are NOT a VT emulator. I do not care how many documented escape sequences you handle or how many *nix applications you work with, until you can telnet into the Deathrow cluster, log in as Demo User, type EDIT A.TXT at the command prompt, hit the </> key on the keypad to bring up keypad help, exit from keypad help then hit <NumLock>-<7> on the keypad to get the editor command prompt, you aren’t even remotely VT compatible. I won’t even go into using <NumLock><*> to bring up the find prompt or the “-” key to delete a line followed by <NumLock><-> to bring it back.

LS-120 drive support comes and goes. I think this has a lot to do with some underlying changes to the kernel and boot loader in general. They are completely re-vamping how they do devices in this release. I do think they will launch this release as is come the October release date, but I also think few will consider this brave new device methodology successful. Nobody seems to be working on the LS-120 issue directly. It appears that the issue is directly related to some other problems though. A good many of us author types continue to use LS-120 and LS-240 drives along with a rotating stack of disks for work-in-process snapshots. It allows us to do off-site storage at friends and family with media that is continually re-usable (i.e. GREEN) rather than one time only CDR or the more expensive few-times-only CDRW. While we have thumb drives, most of us have processes in place, along with storage cases, for the LS-120, so we continue to use it.

The shortage of development funds is really apparent with OO 3.1. The file save dialog has blank check boxes and doesn’t actually add default file extensions, even if you have that option selected. It also cannot import any of the normal supported image file formats, most likely a result of the file dialogs being so broken. Both JPEG and PNG give the error “unknown graphic format”. Another killer for me is the breaking of the Weblog plug-in. One of the few reasons I wills start OO instead of Symphony these days is to write a blog entry. I think I know what the developers broke, I just don’t know if they can fix it. OO only works with Mozilla. This version of KUbuntu had a new default browser named Arora introduced. Let’s just say that Arora was a nice idea that didn’t come close to working out. Konqueror was magically set as the default browser after one round of patches during this alpha/beta process. A lot of sites don’t work correctly with Konqueror so I changed my default browser to Opera as will most people who do any significant amount of surfing. The choice didn’t really matter for this particular problem. OO 3.x only works with Mozilla. This release attempts to use the system default browser. While the upload will eventually get to your blog, the process will die on your computer with various errors relating to the browser. Prior versions of OO forced the user of FireFox with this plug-in, and perhaps OO in general, not so in this release.

I’m sure a large team is working on these bugs. Indeed, there is yet another OO release waiting for me in my updates list. Updates are occuring at a furious pace now that the freeze is on. It is nothing to run an update in the evening and find over 100Meg of updates waiting for you at some point the next day.

DO NOT INSTALL THIS RELEASE IF YOU HAVE MORE THAN ONE SATA DRIVE AND MORE THAN ONE OPERATING SYSTEM.

I cannot make that warning any clearer. Prior to joining this beta, I wiped my entire machine. I trimmed the RAM back to 4GB and used some bootable CD-ROM based utility I had to blast the drive contents and install fresh MBRs on each drive. My machine has 250GB, 1TB, and 1TB drives installed. They are viewed in the BIOS that same way with no remapping. All drives are SATA. I almost wish I had a 500GB SATA to use for an experiment, but I don’t at this time.

There were a few DOS things I was going to need to do, so before installing the alpha/beta, I installed FreeDOS. The 250 is seen as the first disk drive and the primary boot device by the BIOS. Sadly, both 1TB drives are not only the same size, but the same brand and model, so other than by partition table, one cannot tell them apart. I know, I know, but I was going to play with some RAID settings at one time to reduce the number of backups I need to make…before I found out the nVidia RAID was “soft” RAID, not actual RAID, and many Linux distros had trouble with it.

FreeDOS installed, I even broke out my old DOS editor VEDIT to do some editing. Just for grins, I installed a DOS based version of WarCraft to see how things worked. All was well, and I was happy.

Then I installed Karmic Koala. I was somewhat befuddled when I went into the custom partitioning and the 250 was listed as /sdc not /sdb, still, I put my boot and swap partitions on the 250, used one of the 1TB drives for /home, and another to store my /db_data and /books trees. The boot and swap were placed at the end of a FAT32 extended partition which I use for transfering data around. I rebooted as was rather pleasantly surprised to see my FreeDOS partition in the Grub menu…at least until I tried booting it. No amount of tweaking can fix this. Grub-pc is mapping the drives in the wrong order. Every DOS and most flavors of Windows respect the BIOS drive mapping order. That is why you can have a BIOS option which allows you to change drive ordering and things work. Grub-pc and the boot loader are not respecting this ordering. You will be okay if you only have one disk drive. Those of us who do a lot tend to have multiples, and that is where the problem lies.

Without swapping out one of my drives, or adding another, I cannot tell if Grub-pc/boot loader is taking the second drive, the last drive, or sorting for the biggest. From a DOS and Windows perspective, it doesn’t matter, you cannot boot from the provided boot manager. You would be better off if they left all other entries off the menu since that would alert you to the problem early on.

Remember that boot loader which came with OS/2 Warp? People bitched that it had to have its own 1.2Meg primary partition, but it was genius. It was optically isolated from everything but physical drive failure. No matter how corrupted one of your bootable partitions became, you could always boot another entry, hopefully one with fix-it tools to fix the corrupted partition. We haven’t had that in a while. PowerQuest had something they sold which was Windows based, as did many others. Problem was, when your Windows boot partition was corrupted, you couldn’t boot anything else. They seemed to lose sight of the fact that most of us had other bootable partitions not just because we hated Microsoft trash, but because we had utility partitions off on other drives which let us do full backups, virus scans, surface scans, drive partitioning and partition repair. No, sticking a bootable System-Rescue CD in and booting that isn’t quite the same thing. (Though I have had to use my System-Rescue CD more now that the freeze went in than I did when the project was in wild west mode.)

This move to be free of the BIOS was a bad idea. I believe part of the reason the developers are trying to move away is the fact that no Unix or Linux distro runs worth a damn on midrange and mainframe computers. Yes, there are a lot of different flavors for those computers, some are even put out by the hardware vendors themselves, but they all outright (&_&*ing suck when compared to the non-Unix/Linux operating systems on those same platforms. The Linux crowd is learning the same thing the PowerBuilder crowd learned. It is the same lesson every PC product foolish enough to pay the Gartner Group to market it learns. PC stuff does not scale up. There’s a reason OS/Z, MVS, and OpenVMS don’t run on x86 CPUs and it has nothing to do with greed. The simple fact is that the x86 was never designed for business use, it just got used by business.

Eliminating support/reliance on the BIOS is one of the many thousands of things which has to be done to get a version of Unix/Linux to come even remotely close to competing with OS/Z, MVS, or OpenVMS. Of course, to achieve the goal of being able to replace those operating systems, you have to abandoned support for the x86. There is no operating system method of bridging that hardware gap.

Intel has tried valiantly to bridge that gap. They got caught red handed thieving Alpha CPU technology from DEC (Digital Equipment Corporation). Rather than do the right thing and put all of the Intel board in prison, GQ Bob cut a deal letting Intel continue to thieve technology from the greatest CPU in production at the time. Intel worked tirelessly, along with HP and millions (if not billions) of R&D dollars to try and get into the 64-bit chip market. Then, it happened…roughly 10 years late to the party, HP and Intel proudly strode into the media spotlight with the first (and possibly only) Itanium CPU which didn’t melt, smoke, or burst into flame when operated at its rated speed. Once they completed development of the fire retardent system, known as the clock speed limiter, they quickly ceased manufacture and sale of the Alpha which easily had more than 5 years left of engineering improvement without requiring any significant breakthroughs. Of course, the chip was quickly nicknamed the Titanic by the industry, but the industry had to concede they had definitely located a middle ground. You could load Unix, Linux, and OpenVMS on the same CPU and none of them would run worth a damn. Ah….the Holy Grail….