I have a Mac laptop, which I am very fond of; and on it I use Quicksilver which I find inordinately useful (to the point that I have trouble using other people’s Macs if they don’t have it). There is a project available entitled Gnome Launchbox which claims to be an effort to do a similar thing for Linux, which I’ve been meaning to experiment with for ages.
First order of business: upgrade to etch, as GLB requires evolution-data-server 1.2. I was expecting this to be a nuisance, & possibly lead to an exciting couple of hours of post-upgrade fixing. However, in fact it all went absolutely smoothly. I am most impressed (not to mention relieved, as I have a whole bunch of machines to upgrade when it finally gets released).
However, gnome-launchbox was still no go; the version of GTK-2.0 required is 2.10, and etch only provides 2.8.20. I’m not sufficiently keen on this exercise to start building from source or hauling in packages from unstable. I tried the 0.1 release (current is 0.2) but that produced weird make errors.
The next possibility was Katapult. Which may or may not be a good piece of software; sadly their webpage times out so I can’t find out.
Anyone encountered anything else that has a similar function? I may just switch window managers (I’ve been recommended xfce) and start setting up lots of keyboard shortcuts…
Not that it’s anything new, I’m just grumpy that this sort of crap is so widespread and so tolerated:
Blog death threats spark debate
“Prominent blogger Kathy Sierra has called on the blogosphere to combat the culture of abuse online.”
“It follows a series of death threats which have forced her to cancel a public appearance and suspend her blog.”
Gee, a tiny bit of the rampant bullshit that’s directed at women online finally makes the news. Maybe now all those idiots who tell us it’s not a problem and we shouldn’t let it bother us with get a clue? Nah, probably not. Idiot reform is a hopeless cause. But dealing with this sort of junk on online forums and blogs isn’t rocket science- it’s called “admins, grow a spine and hit the delete key.”
Last month Eric S. Raymond made a public announcement on the Fedora developer’s list that he was giving up on Fedora Core and that from now on Ubuntu is his distribution of choice. Actually it was more of a rant than an announcement. ESR’s scatter shot attack on Fedora was wrong in more ways than I care to comment about here. Chromatic did a nice job of attacking the rant on several key points. He also pointed out quite correctly that ESR’s accomplishments as an Open Source activist didn’t make his changing distributions newsworthy.
The reason I’m not simply ignoring ESR is that he did point out a real problem with Fedora Core, one which I also noted in my review of Fedora Core 5 last year. When upgrading a system with yum or pup I ended up with upgrades that couldn’t be run because of dependency issues. ESR ran into a more serious example of the same problem: an automated upgrade breaking significant parts of his system due to similar dependency issues. While the Fedora team at Red Hat has generally been good about fixing such issues within a few days it’s really disappointing that they are still happening and, in fact, becoming more serious with time.
ESR blames rpm, which he believes to be buggy, and yum, which he believes to be overly complex. He is totally wrong on both counts. rpm is a stable, mature, and excellent package management system used by more Linux distributions than any other. yum is no more complex or difficult to use than Debian’s apt and has very similar functionality. Nope, the problem isn’t in the code. That’s the disturbing part of the story, the real piece of news, that everyone seems to be ignoring.
A couple of times I’ve had users complaining about cursor “ghosting” problems while running Gnome - that if typing something, the cursor seems to rewrite itself as it normally would (to move along the line) but not to delete the old cursor positions. So you wind up with a line of cursors. Admittedly this is not disastrous but it is irritating.
I’ve only observed this on systems with the following features: running Debian etch (testing); NVida graphics card; using the nv driver in xorg.conf.
The solution which has worked for me is simply to download and build the nvidia driver (see here for a quick how-to), edit xorg.conf to use that instead of nv, and restart X.
If you’re still seeing the problem after restarting X, check the logs (/var/log/Xorg.0.log) to make sure that the nvidia driver really is being used. I once found that it wasn’t; ps -A revealed a long-running X process which had to be killed before the new xorg.conf settings were picked up.
I have a 1TB RAID array that has to be moved from my one remaining Solaris machine (which will no longer boot) to a Linux machine. I was expecting that - as with the other disk which has undergone the same process - this would appear as a SunOS usr disk and partition, which I could then mount read-only and dump elsewhere before reformatting the disk as ext3 & dumping it back again.
Not So.
fdisk -l gives me:
Disk /dev/sdc: 983.5 GB, 983542530048 bytes
256 heads, 63 sectors/track, 119108 cylinders
Units = cylinders of 16128 * 512 = 8257536 bytes
Device Boot Start End Blocks Id System
/dev/sdc1 1 119109 960490751+ ee EFI GPT
This was a surprise as I had to reformat this disk about a year ago & I do not recall doing anything peculiar to it. EFI, from an initial poke around, seemed to be an Intel PC disk format, which was further surprising since, as you may recall from the first para, this disk came from a Solaris box. However, apparently Solaris 10 uses EFI labels on disks by default, which sounds like a probable/plausible explanation.
The next and more important question: am I going to be able to access this data from the Linux box? Initial answer seems to be no (see the discussion of platforms near the end); there is however a FUSE port - although presently only in beta.
Conclusion: first, another go at resurrecting the extant Solaris box; next, dig into FUSE. Other thoughts, experiences of FUSE/ZFS, or options I’ve missed are welcome…
(For anyone out there muttering about backups: yes, this data is in theory backed up. However, restoring it is proving a challenge as Bacula keeps falling over while it’s creating the file tree. I’m not keen to do the wipe-reformat-restore thing unless I’m confident that the restore will work and at present I’m not.)
Qwaq announced a secure virtual workspace product, called Qwaq Forums, yesterday. It runs on recent Linux, Windows and Mac computers. The product signifies an important milestone for the open-source Croquet project and Squeak. Although Squeak is already the foundation for many great applications (Etoys, Etoys on OLPC, Seaside web server, the Sophie multimedia document creator, to name a few), this announcement catapults Croquet quickly into the business realm.
What is a Qwaq Forum? Unfortunately, I can't give you first hand report on Qwaq Forums. Hopefully, someone will place a video demo online for us to view. I can tell you about Croquet, though. It’s an open source 3D development environment to create distributed multi-user virtual 3D applications. It's quite a freeing feeling walking, talking and collaborating with others and directly with applications floating in air. Anyone can download Croquet and try it out. Please do, you'll have fun.
Back to Qwaq Forums. The site says that Qwaq Forums are "virtual spaces for real work." I imagine a room where the Forum owner can maintain the security level of the room and of each person entering (or not) - from individual rights all the way to anonymous users. Picture a secure office that can be locked up at night with a security card reader on the door - but you don't have to physically travel to use the room! Being a user of Croquet, I can easily see how the room would work. Besides what I mentioned previously, you can also write code and change your environment in Croquet - it's a totally open platform.
An important feature that Qwaq cites is persistence: "all users can see all previous changes and additions" in the Forums and teams can maintain their work and the progress made. From the datasheet: "Setting up a Qwaq Forum is simple: start by simply dragging and dropping content into a workspace in Qwaq Forums. All otgher users present will see the content immediately and will be able to start working with it right away." I don't know what could be easier.
I won't know more about the Forums until I get my hands on it. If a video becomes available, let me know. In the meantime, check out Croquet and Squeak and see what you can do with it - your application in this 3D space could literally change the world. The field, uh.. virtual space, is wide open!
In my last article I cited the Vector Linux developers as an excellent example of the way Open Source developers respond to the user community. All of us who benefit from Linux and/or the myriad of Open Source applications out there are part of that community.
Tuomo Valkonen, who you may know as the author of the Ion window manager has a rant about the brokenness of the megafreeze development model, where a Linux distribution attempts to stabilize the entire system.
As an upstream contributor, I find it highly frustrating to release a new version of my software and have users contact me for about ancient versions I no longer want to support.
Perhaps if every project moved to a monthly stable release cycle then it would be easier to create an entire distribution that’s not all out of date. As it is, trying to wait until the stars and planets and comets and asteroids all align for one blessed week of complete stability… well, even I don’t believe in trying to herd those cats.
Alternately, stabilizing only the base operating system–the kernel and the core utilities necessary to install further software packages and bundles–might be sufficient. Perhaps the real problem behind this big problem is that the problem is just too big. Maybe a monolithic release is just too much to organize all at once.
I missed this in January, but as I look at Penguin.SWF and Adobe Flash Download Center, I see that the Linux versions now have the x86 label.
While it would be nice to have an open source Flash player (so that all platforms and architectures could use Flash–meanwhile Gnash continues), I do appreciate Adobe labeling the player appropriately. Saying “Linux player (x86)” makes it much clearer what Adobe currently supports.
As of this morning, I found that ssh logins into my Debian etch boxes were monumentally slow. Using the -vvv switch it looked like the problem was down to a very long wait for gssapi-with-mic authentication. Trying with the -o GSSAPIAuthentication=no switch on the command line helped, although some boxes were also pausing for a very long time when dealing with public-key auth.
Upgrading to the unstable package of openssh-server seems to have fixed the problem; although I can’t find anything explicit in the bug reports. Since libkrb53 was upgraded recently I’m wondering whether the source of the issue is actually in there.
However, note that if you’ve previously had ssh-krb5 installed, and have had the upgrade to openssh-server occur, you may have an old ssh-krb5 process kicking around. I found that in general, /etc/init.d/ssh-krb5 didn’t get removed; /etc/init.d/ssh just got installed alongside it, so be careful when rebooting.
Dear Mr. Shuttleworth,
Get into the hardware business.
I saw you speak in 2004 in Gothenburg Sweden at the EuroPython conference, you gave the keynote (excellent speech) and told the audience about your upcoming project. It was a linux distribution that would be focused on the desktop and come with commercial support, two things you saw as being necessary to linux success.
Now Canonical and Ubuntu have made great strides toward realizing your vision yet there remains a bottleneck - people still have to install linux. Installing linux is easier than ever, but installing any operating system is unfortunately not trivial and is a chore that most computer users have no use for and will avoid. Remove this impediment, sell linux pre-installed.
Your company has been building a support network to support Ubuntu, this is a key ingredient in any successful hardware offering. Lack of a support system for linux is cited by Dell as a reason why it cannot just open the floodgates and soak the masses with cheap laptops. But you have that support network already built.
There are partners out there to potentially work with, like Asus or Acer. I think Acer would love to build Ubuntu branded laptops, then they could really challenge Dell. This is of course speculation on my part but the 100,000+ votes for pre-installed linux on Dell’s idea storm shows that there is a business case to be made for a Ubuntu laptop. You can even go upmarket and get IDEO to design your laptop or desktop, there is still a decent profit margin there and free software tools for creatives like Inkscape and the GIMP have improved substantially which might bring some of those taste makers onboard.
In short;
- there is a market for pre-installed linux
- you already have the support system
- potential partners exist in the hardware branch
- you already have the software
Thanks for reading Mark and I hope to be able to buy a Ubuntu laptop in 2008, put me down as your first customer.
Here’s a frightening quote from GNOME’s Bugzilla:
I think open source inherently creates a much more useful
Novice/Expert barrier anyway. The real experts can do whatever they want,
because they have the code, and can change it to their liking.
— comment #13 in configurable mouse click actions patches
Is it really better to encourage “experts” to maintain local code forks than to add configuration options? If the defaults aren’t right for you, you’d better hope you’re an expert and can change things to your way of working–just don’t expect to get that patch upstream so other people can benefit from it. If you’re a novice, you’re in trouble.
That’s… not exactly my understanding of the point of open and collaborative development. I hope it’s not a widely-held opinion.
In January I reviewed Vector Linux 5.8. While the review was mainly positive I did complain about what I saw as some faults in the distribution. The response from the developers of Vector Linux was almost immediate, both in the Vector Linux forum and in the comments under my reviews, and was incredibly positive. In the weeks since then a surprising number of changes and improvements have already been made, particularly in the area of internationalization and localization. A suggestion I made in the VL forum regarding compiling and building packages to insure that localization files (translations into various languages) are included was taken to heart by the developers. The result is that when an updated Xfce package was built after version 4.4.0 was released it included the additional language support. So… if you’re native language is Danish or Hebrew or any other language for which translations exist in Xfce you’ll find, for example, suitable menus for your desktop.
While this may sound like an exceptional example of accepting and acting on feedback from a user of a particular piece of Open Source software it is really, in my experience, something that happens all the time in the Free and Open Source Software community. This is why the pace of development of Windows or most commercial UNIX flavors seems glacial to Linux users and professionals. We’re used to a level of responsiveness from developers that most proprietary software vendors simply cannot match.
I’ve used both Windows and Mac OS X, and I don’t know how people manage to install software on them. Yes, I’m a fan of aptitude (and I’ve had good experiences with Yum).
It’s pretty clear to me that this software installation system has tremendous advantages over the traditional “download a random binary package from somewhere” approach common to Windows and Mac OS X, as long as the package you want is in the proper repository.
I have posted along these lines before.
This time the recalcitrant machine is a SunBlade 100. I had at some point in the past managed to get it as far as a Debian testing install, but then it wouldn’t reboot (hung at various stages during the process). A correspondent who’d read the previous post suggested that I try appending ide=nodma to the boot line (so in my case: Linux ide=nodma at the SILO prompt). Worked like a charm. Splendid.
Unfortunately, I then messed up my /etc/silo.conf trying to add this to the boot line so it would Just Work rather than requiring manual intervention at boot. Messed it up so badly that it - once again - won’t boot at all. (One of the morals of this story is: always run /sbin/silo after making changes to /etc/silo.conf - something which I had firmly in mind back in the days when I used LILO, but which has slipped out of awareness with the ascendence of GRUB.)
Even worse: reinstall won’t work either. Both Ubuntu and Debian etch get as far as the “Booting Linux” line and then hang silently (in either install or rescue mode). I’ve tried the various fixes on the previous post, but no dice. Any further suggestions gratefully received! It is a lovely big paperweight, mind.
Notes from my last post: httrack was just the wiki backup I required, after I told it to ignore all pages including the word Special (i.e. all metadata / change record pages) as there were just too many of them and I don’t need them for emergency purposes.
httrack http://server.com/wiki/ -*Special*
does the job. Thanks to William for the comment!
