Older blog entries for ken (starting at number 2)

The release of Knoppix 3.4 was something that I had been eagerly awaiting for one major reason - kernel 2.6 on a pre-packaged distribution. I needed the kernel because the old 2.4 series did not support my laptop's troublesome chipset all that well. This entry covers the process and challenges involved in getting a Linux desktop to run well on my laptop.

Laptop specs:

First off - knoppix wouldn't boot on my laptop at all. After the initial knoppix screen, it would boot into a blank screen and freeze. All the standard options, such as failsafe, expert, etc, all failed. An initial search suggested that I needed to set the "noapic" option, but even that did not work. After much searching and comparing, the magic sequence to boot knoppix on this laptop turned out to be:

knoppix26 noapic nolapic noscsi

Yes, scsi probing causes the system to crash. The "no local apic" option was also something that needed a bit of searching for. I found it in the kernel-parameters.txt file in the source tarball. This managed to successfully boot knoppix though, and all worked well from then on.

The next problem then proved to be actually installing it on my laptop, since like all WinXP preloaded laptops, it came with WinXP taking up the entire hard disk and partitioning was required. Knoppix 3.4 came with several partitioning tools (QTParted, ntfsresize) but the versions were relatively old and unable to deal with WinXP's peculiar usage of disk space.

WinXP (and probably all other NTFS OSes) tends to leave a large chunk of data at the end of the partition allocated to it, marking it hard to resize the partition. Running the included Windows defragmentation tool does not budge this large chunk at all, even if the partition is 90% empty.

The versions of QTParted and ntfsresize on Knoppix 3.4 were unable to shift this chunk so it would have required a destructive repartitioning of my laptop. I wasn't keen on reinstalling WinXP (no, I still need it on my laptop) so it seemed like a bad situation. Fortunately, the newest version of ntfsresize (1.9.1) is able to move those chunks of data. I downloaded the ntfsresize tarball onto the NTFS partition, booted Knoppix 3.4 from the CD and unpacked the tarball into /home/knoppix (don't unpack into the NTFS partition! The NTFS drivers on Knoppix 3.4 are still beta and can thrash your NTFS partition) and ran it from there. The partitioning went off without a hitch.

After that, knoppix was installed onto the laptop using knx-install (verify). Being a debian user, I was quite pleased that it offered a "debian" installation. Didn't really bother to find out the differences between the different installs but the bootup screens swapped to the debian sarge (unstable) logos instead of knoppix.

Only problem is that the install program didn't give me an option to pass kernel arguments, so the laptop was still not functional on bootup. Despite using lilo as the bootstrap loader, it wasn't possible to pass arguments on startup. It was back to the knoppix 3.4 CD one last time ...

Running from the CD this time, I mounted the HD partition and chroot'ed to it. After editing /etc/lilo.conf to pass the arguments (and making kernel 2.6 the default kernel), I updated the bootloader and all was good. I had a debian system running on my laptop and all was well ... until I wanted to use my wireless card. But that's an issue for another day and another blog.

4 May 2004 (updated 4 May 2004 at 16:36 UTC) »

Two new announcements today that I found interesting. First, Knoppix 3.4 is out! For those who aren't familiar with it, Knoppix is an excellent live-CD that you can pop into a CDROM drive, boot from it and experience a full GNU/Linux system without any hassles, installation or changes to your computer. Useful for a variety of functions, including introducing newbies to GNU/Linux, recovery CD or just plain portable linux on a CD.

Personally, I've been waiting for Knoppix 3.4 because my laptop has an unusual chipset (Packard Bell Easynote) that requires a very late 2.4.x or 2.6.x kernel above to support it. Knoppix 3.4 comes with both, so I can finally put a worthwhile distribution on my laptop. It was a race between Fedora Core 2 and Knoppix, but it looks like Knoppix got there first. Still awaiting the bittorrent download to finish though :) Other fun goodies on this release include KDE 3.2, OpenOffice 1.1.1 and Gimp 2.0.

The other interesting release is the Red Hat Desktop. In a reversal from their abandonment of Red Hat 9, RH is now back on the desktop. Among the more interesting details of this deal is that they are bundling a variety of non-Free addins, such as Acrobat Reader, Macromedia's Flash plugin, a Citrix client and Real Player. They have done a lot of relationship building to ensure that a lot of what users take for granted on a windows desktop is also available on their desktop.

It's not cheap though and obviously targeted at the enterprise or really wealthy SMB. The smallest lot you can purchase is 10 at a time, for a tiny fee of US$2,500. That comes to about $250 per desktop! Phone and Internet support is included, but limited. Phone support is for one month and Internet support for just a year. Not a lot at all. Purchases in larger sets are somewhat more affordable. A 50 set expansion set (have to have purchased one of the more expensive options earlier) costs $3,500 for 50 desktops. US$70 per desktop, which is somewhat lower, but still steep for a one year license. Windows XP is cheaper, especially when amortized over 3 years.

22 Apr 2004 (updated 26 Apr 2004 at 07:07 UTC) »
FOSS will lose in developing countries

FOSS will lose in developing countries and proprietary software will dominate these countries even more strongly than it currently dominates mature economies such as the US and Europe.

Why do I say this? Aren't tons of developing countries drafting or implementing open source policies? Haven't there been a slew of positive reports on FOSS from development agencies and NGOS? Doesn't FOSS provide a huge number of compelling benefits for developing nations?

Yes to all of those questions. They don't matter though, because ultimately all of the above are just words, not action. This is action:

Microsoft announces US$1 billion dollar partnership with UNDP to assist developing nations.

The last community related announcement I have seen from a FOSS company is this one:

RH launches US$23,000 scholarship programme in India

Not to belittle Red Hat's efforts or anything. They are doing something really admirable there. But what policy maker will look at the two headlines and say, "oh, let's go with the US$23k dude". These announcements make a big difference. Within days of the announcement, policy makers were going, "crap, let's reconsider these FOSS projects, otherwise we won't get a cut of that US$1 billion." Got to hand it to MS - they know good PR.

Where are all the FOSS companies? Have we ever seen IBM announce a global partnership to assist countries? HP? Sun? Any of the other major FOSS companies? Nope. Would love it if someone would point them out to me. All I see are the announcements of IBM signing their latest deal with some government, or Sun, or HP. All business transactions. All profits for these companies, no cost.

"But they don't have US$40 billion sitting in their bank account like Microsoft!" the FOSS defenders will cry. Big deal. You don't need US$1 billion to ante up at this poker table. Microsoft isn't throwing US$1 billion in cash in their "Unlimited Potential" program. Check the details carefully. A little bird (homage to Jeff Ooi) told me that MS isn't funding hardware purchases. So that leaves mostly software and training, valued at market rates. How much does it really cost them?

"IBM and Red Hat can't do that! Software is already given away free and trainers still cost money!", more FOSS defenders will cry. Right. IBM and Red Hat have other valuable resources that developing nations desperately need. Can't give away training? How about training materials? How about projects to start self-sufficient training centers in a country? Doesn't take much - send in a few trainers with your training material, train a few more trainers there and the cycle goes on. Don't want to canibalize your profitable training business in developed countries? Pull a Microsoft. Give away material in a non-English language.

Bottom line? FOSS companies need to step up to the plate and be seen as doing more to help developing countries. It doesn't have to cost much. Talk to some NGOs that are actually on the ground. They know how to do much with very little. All you need is a little creativity.

Not that it will make them any sales or profits in the short run. But in the long run, would you rather be competing in a market where your competitors dominate 95% of the market or 50%? Are you going to abandon the future for short term sales today?

New Advogato Features

New HTML Parser: The long-awaited libxml2 based HTML parser code is live. It needs further work but already handles most markup better than the original parser.

Keep up with the latest Advogato features by reading the Advogato status blog.

If you're a C programmer with some spare time, take a look at the mod_virgule project page and help us with one of the tasks on the ToDo list!