A view from the hill

Dell is selling Ubuntu equipped systems since about a year now and seems to be quite happy with it. Whatever that effectively means, at least I can tell that I'm quite happy with Linux on Dell systems, too.

Through the last five years, I've been using Linux on a number of Dell systems. Under my personal control there have been three laptops (Dell C610, D610 and a Latitude 640) and a desktop (Optiplex 755), on which I have been running Debian Sarge, Ubuntu Dapper, Feisty and now Hardy. We also had several Dell servers at work running more or less smoothly with Debian (sarge, etch). Using Linux wasn't always without problems: I had trouble with built-in modems, PCMCIA ISDN cards and acpi/hibernation. For example, on my private Latitude 640, I have trouble suspending at all, because of the ipw3945 driver for the wlan. But the important thing to note is that basically all problems were really small and never of a size requiring me to use some other OS in the first place.

The only real issue is not with Dell per se, but more with my favourite OS, Debian: over the years, and especially with the ever-lasting sarge release, getting Debian to run on a recent system got more and more difficult. That's the main reason why I've been using Ubuntu on all recent hardware I had contact with: it's more or less (more so than less) a Debian system but does run on modern hardware. Main issues here were graphics adapters, sata/scsi hostadapters and network/wifi cards, or to put it otherwise: too old kernels, too old X.org. Both problem sources can simply be solved by using a recent version of Ubuntu. Sorry, Debian, but your release cycle is just too long to be acceptable. Granted, all these problems are mostly an issue when installing a new system, but it's not always possible to plug in some old disc with a working version of Linux.

Recent fun:
- Working hibernate/suspend without any manual configuration on a brand new Dell desktop using Ubuntu 8.04 (beta).
- Watching a self-generated video CD (probably done under Windows) under Linux without any problem with my wife that we couldn't watch on Windows due to scrambled colours under Windows media player.

Recent less fun:
- Hibernate/suspend on our older Dell laptop with Ubuntu 7.10 and finding it's a known issue with the ipw3945 driver.
- Getting vmware to work on a recent 2.6.24 system.
- Having fixed that finding out the hard way that Aero won't display in the VM, requiring to go back to cloning the (not so small) physical machine a third time.
- Being unable to get a DHCP lease over the rt2x00 driver for my USB wlan stick with kernel 2.6.24, despite being able to get a sucessful connect via wpa_supplicant/wext.

Two days ago, I updated my laptop from Ubuntu 6.06 "Dapper Drake" (aka LTS) to "Feisty Fawn", aka version 7.04. Just in case you wonder, how I did that, I just followed the usual directions and first updated to Edgy and then to Feisty via issuing "gksu 'update-manager -c'" twice. Things went quite smoothly, but I did encounter some problems:

- I lost access to my databases during the migration from Postgres 7.4 to 8.2., i.e. there was no proper migration and as support for version 7.4 has been dropped, I even have no way to dump the old data and import it again. I need to figure out whether I can easily access the old dbs from Postgres 8.2 (mind you, I'm talking of the raw data and the DB server, not client access).

- As a result, my local copy of serendipity stopped working. I re-installed serendipity (a clean start seemed to be a good idea), but still Iceweasel gets wrong information about the phtml files it generates. Probably a problem with the PHP installation, which I need to get straight (I'm not a PHP guy, though).

- When booting into Edgy, when switching from the splash screen to X. the screen went black and there was no way to get a display. I solved that by ssh-ing into the box, modifying /etc/X11/xorg.conf by using the ati driver instead of the fglrx driver and then editing /boot/grub/menu.lst so that it wouldn't show the splash screen or suppress messages. The problem went away after going to Feisty.

- To get VMware running again, it wasn't enough to issue the usual /opt/bin/vmware-config.pl that you have to do with every new kernel. It kept telling me that the headers directory wouldn't fit my current running kernel. Fortunately, I already had loaded the new version down which solves the problem (I guess there is a new version numbering scheme in newer kernels).

- The update of course also brought with it the switch to Iceweasel, so of course, I lost some of my trusted extensions (Reveal, Surfkeys).

- What really annoyed me was the ugly, fat looking "optimal" font that I encountered in most applications. Why Gnome doesn't use subpixel hinting as a default is beyond me (especially as I'm quite certain to have had that setting in the old setup, too).

- The new eclipse version of course knew nothing about my local installed plugins, implying that I had to reinstall them.

On the nice side, my mail setup kept working and the network setup kept working, which I think is way more important.

Etch ships with drivers for the Ralink RT2570 wifi/wlan chipset in the package rt2570-source. However, for whatever reason, if I try to use that package (i.e. compile it the usual Debian way via make-kpkg modules-image and install the resulting deb), modprobe rt2570, I only see that the usb subsystem sees the adapter but I get no interface (rausb0). I then just recompiled my local CVS copy from last summer, modprobe/insmod it and everything is fine. Huh?

Update: I updated to 2.6.21.1 today, using a fresh vanilla kernel. Still Debians module compiles flawlessly, can be installed and the module installed, but no working interface shows up. Manual compiling/installing the old CVS version (probably by a month younger) works out of the box.

In my workstation I still operate a Matrox G400 back from 1999. It was a state of the art graphics card for its time, and especially well supported under Linux/XFree86. Well, time went on and now everybody's using X.org. Since two days ago, my workstation joined the club, finally having been updated to etch. And then I find a discussion in the Usenet group de.comp.os.unix.linux.hardware with somebody complaining about a freezing system with X.org/mga when switching to the framebuffer console. And of course, I can easily reproduce it: Switching back to X immediately hang the system. Suspend to disk hangs up the system, too. Great.

A little bit of searching quickly revealed this horrible list of bugs for xserver-xorg-video-mga in Debian, from which one bug in particular links to this upstream bug entry. See the date of the last change to this bug? It's almost three years old! I am now trying with "UseFBDev" set to "false" and at least I could switch three times back and forth, but we will see.

I want that keyboard, really. Now, if He would lend me one right hand for one of my two left ones ...

No, I'm not talking about the lasershow Tool were playing their music to.

Yesterday, I finally got my new printer, a Lexmark E240. I was pretty certain that it would work, as the E120n is known to work as a Generic PCL6 printer and Lexmarks specs for the E240 claimed it understood PCL6 as well. So I just tried the driver (pxlmono) that's recommended for the E120n and it worked right out of the box (the entry for the E240 is from me). I believe that the E240d(n) as well as the E340 series are likely to be the same as the documentation coming with the E240 applies at least in part to those as well.

Lexmark does provide it's own set of drivers, btw. They ship (at least on the supplied CDROM) as RPMs, although they claim to support Debian. It looks like they provide some administration tool, and from a quick sketch in the documentation, I think they mainly provide a wrapper around the usual printing tools. But I haven't tried this stuff at all. Speaking of documentation, the Unix docs ship as HTML, and at least in part they are broken. I.e., the main navigation frame links to some HowDoI.html file, but on the CDROM there's only howdoi.html. Let's hope the hardware engineers paid a tiny bit more attention to what they were doing.

Peter vanEynde discusses his experience with binary-only drivers for his ATI card:

The machine (a Lenovo Z61m) uses a X1400 ATI chip, so I have to use the ATI binary-only driver. Well. I'm trying to use it. The amount of grief I'm getting from it would convince even billg himself that opensource drivers are mandatory in an opensource system. Strange errors pop up and you're just lost, googling only gives you forums where other poor users are trying to make it all work.

I think he's wrong about convincing that billg. Isn't that just the current Windows experience brought to Linux?

FWIW, when I needed a new laptop for work, I also ordered one with an ATI chip. There is a working X.org driver for it, though, and I double-checked prior to the order. On the other hand, I'm currently using the Ubuntu provided proprietary ATI driver without any trouble. But even if ATI decides in half a year that it no longer wants to support my ATI M300 with their linux driver, the open source driver will be there. So yes, I'm all for open source drivers, but I would be happy with a scheme that has been applied by several vendors in the past: keep the driver proprietary for some time, but release it as open source when your technology has moved to new heights. Perhaps Nvidias and ATIs current behaviour just reflects that there hasn't been any advances in their drivers?

It's no fun when the only thing you can't check in a virtual machine test installation is the connection thingie you have to use when on the road. In my case, the AVM Fritz capi module shipped with Ubuntu Dapper Drake is broken. The bug report is about two months old and still has status "unconfirmed"? At least the bug is not in the binary only firmware but in the source around it. I knew there is a reason I like open source, even if this "open" tastes more like "open beer" in this case (the Fritz drivers still only target the ancient SuSE 9.3).