Doing the Monster Mash

This is another one of those posts where I am going to use the term 'Linux' to describe a whole stack of projects. I know Linux is really only the kernel, but a kernel without the rest of the projects that orbit it would not be very useful. At least, not as a desktop / laptop OS. I suppose you don't need much more for embedded OS applications. Depending. Its the first paragraph, and I already digress..

Work has been crazy busy lately. So much so that I have not yet really spent as much time as I should have doing the hard work of getting ready for LinuxWorld in the evenings. Or writing this weblog. Part of it was that I was writing an article about picking a Linux laptop for the May issue of Linux+DVD.

Help: I can't stop trying new things!

Another major part of it is that I can't seem to stop experimenting with things. I know people who leave the office and don't want to even look at a computer till they next need to. I am like that sometimes too. But quite often not being at work is a chance for me to learn new things about the things that interest me that I was not able to do at office. These are things I take with me to LinuxWorld and SHARE to talk about with others, and I have yet to do an experiment that did not later have a useful purpose of some sort. Knowledge is always useful.

It is also handy that my day job is about the practical application of computer knowledge. Problems at the office suggest experiments at home. Experiments at home present solutions to problems at the office. Looking back at when I converted to a Linux desktop full time at the office, I could not have done that all those years and virus storms ago without know it was possible from things I had learned from experiments before. Idle knowledge acquisition became mission critical when it was the only way I could stay productive at the office.

Open Source and Standards and Interoperability, or People are People...

I have been thinking about and therefore writing a great deal about the wide open topic of open source lately. One of the general sub-topics I was thinking about was program compatibility. Before you even get to standards and standards bodies, open source is, in a manner of thinking, open standards. When one group writes something, anyone who can read the code knows how the program works. Further, there are all sorts of interoperability open standards in place in the Linux ecosystem. Example: The new Portland Project will be a nice step along the way for all Linux desktops like KDE and Gnome to better interoperate.

Then there is the idea that one project can lead to another, and so therefore probably maintains some level of compatibility with the parent. In the extreme case of that, CentOS works very very hard to be compatible with Enterprise RedHat versions. This idea, like the human beings that underly it is not perfect: there was a bit of drama a while back in the Fedora project where some repositories were created to add packages in to Fedora that were being left out due to the way the project was set up. As I understand it, there were some in the Fedora project recommending against enabling these extra-Fedora repositories because the project would not guarantee that the extra packages they contained and added would either be compatible, or at the very least remain compatible with updates that the project was releasing. I assume there was a package that was released or updated in some way that lead to the whole blow up. I don't know. I didn't really dig into it as deeply as I could have. I was not having any problems with anything, so it was only of mild interest. That there were hurt feelings on both sides was clear.

There is a Debian in my Pocket

Another project that has had its share of drama is Debian. As pure a project as one could imagine, it is also often criticized for the glacial pace at which official new releases come down the road. Debian 4.0 was released on April 8^th, 2007, after much teeth gnashing and other internal drama for too complicated to go into here, even if I knew the whole story. Like the Linux Kernel itself, the whole thing seems too big for any one person to understand it all.

Many of the packages in Ubuntu are sourced out of the Debian unstable or experimental trees, which at least in part is probably meant to point out that since Ubuntu is very stable that Debian is being far too conservative with it's stable releases.

Whatever the case is, one can not escape the gravity or perhaps even better, the gravitas of the Debian project.

Any distro that has at its core Debian has a unifying set of over 20,000 packages. It is the largest package set of any Linux I have ever worked with. Last I looked at Fedora for example it was about 25% that large in terms of total packages. Not counting the extra repositories, like Dag or FreshRPMS, etc.

The Debian-based Mash

My new Linux laptop is an Acer 5610. In it's short life with me it has already run many different combinations of things. This last stream was the most interesting to date. I started with Mint 2.2. Mint 2.2 is a derivative of Ubuntu 6.10, with nicer-looking-to-me themes and some handy tools like MintDisk and MintWifi.

On top of that, I installed the KDE packages from the Kubuntu 6.10 repositories. And Beryl for 3D desktop.

Mash 1.0

The point of that technical stop was that the Acer is a dual boot laptop. I wanted to see how MS Vista compared to Beryl/KDE on the exact same hardware. It should come as no shock that Beryl is faster, uses less resources, and has so many configuration options that I'll never try them all. It is an amazing piece of work. I read an article that said Beryl was a fork of Compiz, and that the two projects are talking about working together. I hope they do.

Beryl is amazing, but I ultimately have chosen to use it less and less frequently. I have no use for flash for flashes sake. An effect has to have meaning: communicate something about the program. OS.X is my benchmark there.

3D and other effects: Flash or meaning?

I am not planning on leaving Beryl behind. Just waiting to have some time to start turning some things off. Make it more OS.X like. Only have the effects on that tell me useful things. My favorite effect is the one where the farther 'back' in the desktop stack of windows a program screen is, the more transparent it is. That tells me something useful. Other than that, I want a desktop to snap. I have no patience for example with menu fade. No meaning. Just showing off. I clicked on a menu open because I wanted something on it. Fading in gives the impression to me not of coolness, but of slowness. To me. I know I am not typical here. Enough people don't like it that even MS Windows has an option to turn the effect off. I always turn off smooth scroll from the same reason.

I know I am not typical in this regard. When I was doing a beta test of Outlook for Microsoft, and they introduced the 'Find' button, part of my feedback was that I was not happy that clicking the button for find meant I had to wait while windows slide around to make room for the find bar. Tap. Tap. Tap. Look out window. Attention wonders. Oh. Good. The screen is ready for me to type what I was wanting to find in it. What was that again? Aw nuts. I described this effect as 'The bands in Outlook are slipping': if you have ever driven a car with an automatic transmission that has loose bands, you'll know what I mean. The Beta team basically thought I was nuts. I guess they worked hard to add that effect. I had succeeded in calling the baby ugly. Even in Outlook 2003 this slip-slide-y thing is still there.

Beryl may have thousands of settings. But at least I can turn off the effects that annoy me, and clearly I am annoyed more easily than most by such things.

Mash 1.1

After experimenting with the Mint/Kubuntu/Beryl combo burrito for a while, I was given the option by Adept to install Kubuntu 7.04 Beta 2 when that became available. I assume I must have enabled a reach-ahead repository in /etc/apt. I did not remember doing that, but there was the offer, on the screen.

I pondered that offer from Adept for a bit. Curiosity won. Mint 2.2 is easy to install again if this totally smokes the laptop. Said 'sure' to the upgrade question.

Adept downloaded and applied over 1600 packages, and after the reboot I was doing the monster mash. I did not expect it to reboot at all, or at least to stop in single user mode and ask me if I was nuts. I was happy to see I was wrong: it came up on the new 2.6.20 kernel, and loaded the new KDE 3.5.6 desktop, and Beryl started... and was even faster than before. VMware Player worked just fine. A new warning appeared: That I was running 'Restricted Source': VMware and the Intel 802.11 card driver were noted. I assured it I was fine with those things.

Also of interest: I appeared Beryl was in the base 7.04 repositories now. I did not have to re-add Beryl. It was there on the system, and it was upgraded.

At this point, I am running (in a way of serially thinking) Debian/Ubuntu 6.10/Mint 2.2/Kubuntu 6.10/Beryl/Ubuntu 7.04/Kubuntu 7.04. And the funny thing here is that I should soon be able to enable the Click N Run warehouse, and install packages that started life on LinSpire / FreeSpire. Also Debian based, but still...

What an amazing ecosystem.

The death and life of Gnome

It is not all peachy. Gnome wouldn't work anymore for some reason. In fact, a few days after I did this mash, a member of my team flopped an Ubuntu 7.04 disk on my desk and asked me if I had tried it yet. I thought about it. Should I get to work on the lab update? This would put both Gnome and KDE back to rights if I installed it over the mash. Wait till after Linuxworld?

Nah. Not yet. I have more to learn. VMware Fusion Beta 3 is out and needs to be looked at too! Next week for sure though for Linuxworld stuff. Knoppix 5.1.1 needs all new screen shots for the doc...