I finally got it together to release a new version of tritium. Probably the most exciting feature in this release is that you can allocate an area of the desktop to be used for a status bar / panel / dock area.
Here is a screenshot of my desktop showing a working gnome-panel at the bottom. Debian packages/sources are available here, and from sourceforge. The git repository is here. The next big task for me will be to add some kind of 'scratchpad' functionality similar to what ion3 has. update: A number of people have pointed out some missing files in the .deb, so I have posted a 0.3.2-1 to http://vireo.org/debian/tritium/ ; thanks for the feedback.

Every now and then I encounter a web page which I'd like to archive for later purposes, e.g. because I can see that it might be related to my research, or simply because I want to keep the information until eternity. I know of three ways to achieve this, but each of them has drawbacks:

• Using wget to download the page. This does not pull any dependencies (although wget could), nor does it stamp the downloaded file with the original URL and time it was downloaded.
• Printing the page to a PDF. Apart from being a cumbersome process, requiring at least five mouseclicks and way more attention than this should take (in Firefox at least), printed webpages don't look the same as on-screen. Even though Firefox stamps printouts with URL and time, this information is truncated if e.g. the URL is too long, and in a printout, you lose all the meta information, as well as links etc.
• Using Firefox's 'Save as' feature. Again, being too cumbersome for my taste, this actually manages to properly archive a webpage with all dependencies. However, it drops a file and a folder on your hard disk, which I find rather annoying. And it also doesn't stamp the downloaded files with neither URL, nor time.

So, dear lazyweb, how do you do it? I am aware that it'll be trivial to hack up a little script which uses wget to obtain page and dependencies, create a file with stamp data and produce a frameset to put this data above the actual file, but before I spend time implementing the wheel, I'd like to make sure it doesn't yet exist. NP: The Flower Kings: Paradox Hotel Update: Michael Stevens pointed me to furl.net, but I need a solution that I can use offline. Part of the reason that I want to archive web pages is because I don't want to come back in a couple of months and find a webpage removed, a server dead, or a site like furl.net discontinued. J rg Jaspert introduced me to Scrapbook, a Firefox extension I'll have to check out. However, even before looking at it: I really want files or directories on the filesystem, so that I can check them into version control and have them archived and replicated. I don't trust any data to Firefox or any of its extensions, or anything under ~/.mozilla. This is also the reason why I Slogger doesn't cut the mustard for me. Matej Cepl wants me to use konqueror to archive them. I don't want to install hundreds of megabytes for this feature, nor can I imagine that this feature will be much different or less cumbersome than Firefox's ability to save pages. I hate the mouse. Thanks for all your comments, regardless! Update: Just found this post about archiving stuff like digg/delicious.

I have just added two new categories to my blog, one is for the most popular posts [1] (as indicated by the number of hits on the permalink pages). The other is for the best posts [2]. My criteria for adding a post to the best-posts list is that it provides some information that is new or some analysis that others do not appear to have performed, that it doesn’t get refuted by someone else (sometimes an idea seems good but someone points out a flaw), and that there is some level of interest in it from readers (based on page hits, comments, and links from other blogs). Both of these categories may be added to posts some days or weeks after they are published. So adding the feeds for them to a syndication configuration might not be a good idea as they will always include posts that are old. I expect that a typical Planet configuration would never display posts from those feeds. I suggest that other people consider adding similar categories to their blogs. It will allow readers who quickly browse your blog to see the posts that you regard as your best content and other bloggers in the same space to see what gets the most hits (which is worth-while if you don’t consider blogging to be a zero sum game). I expect that someone will suggest that I only write posts that are eligible for the best-posts category. However this is one example of a post which I don’t consider to be eligible but which will still be useful to some people.

• [1] http://etbe.coker.com.au/category/most-popular/


• [2] http://etbe.coker.com.au/category/best-posts/

Share This

Horray! NVIDIA just released 100.14.19 that is Xorg 7.3 compatible and with a rather big changelog. That should make a couple of Debian users happy. I'll work on packages ASAP. Hopefully legacy versions will be updated soon as well

I am starting to wonder why every single bit of hardware that deals with wireless networks is utter crap. Really, I'd pay good money for an access point that doesn't crash when the firmware on the ipw2200 sends out a malformed packet while crashing because I used all the transmit buffers in the card. The zd1201 is no better, I've managed to have its firmware crash on me twice now, and unlike the ipw2200 which will at least admit that there has been a firmware error and reboot the card it just silently retries for an infinite time.

Lately I have been wanting to buy a lot of fairly expensive stuff that I don’t really need. It started with the telephoto zoom lens, which I ended up getting, and love. I’m really happy that I got it, especially since Lisa and I will be going on a hot air balloon ride in Napa for our anniversary, and then to the Salinas Air Show the following weekend. Then it was the new TiVo HD. We already have a Series 3, but we’ve occasionally thought that we might like to get a second TiVo for our other TV. Now that it’s been confirmed that Multi-Room Viewing is coming to both the Series3 and TiVo HD soon, there’s even more incentive to get one. Fortunately right now they’re selling them to employees at a discounted price (limit one per employee). And now it’s the iPhone. I can’t explain why I feel compelled to buy this other than iWant. A few of my coworkers have one and they look pretty slick. Plus my current iPod battery can’t hold a charge very long. Plus I like the idea of having an integrated iPod and phone (it’s part of the reason why I got my current phone, the Motorola SLVR L7). Plus it’s roughly the same size as the SLVR and only slightly heavier.

I just received the autumn 2007 edition of the Co-operative Group's membership magazine. It came with a cover letter that finally prompted me to answer the food ethical policy consultation The announcement was covered in the Guardian under the slightly inaccurate headline How green do you want your bananas? Co-op ballots members on ethical issues and in the Metro under the headline "Flight logo row" focusing on the criticism of the Tesco and M+S branding (which I think was a little unfair in some ways). Even so, it's good to see the Co-operative Group getting our message across and wrong-footing the BBC's attack dogs, described in Cambridge Co-operator: Ethical Policy. Last time on five live, our man got a bit mauled. Will this change make a difference, like it has to the Co-operative Bank? Can an ethical policy persuade you to change where you buy your food? (2007-09-19: 1 comment)

The other day, as I was putting a couple of pages of documentation together for a company-internal development project, I got this terrible feeling of malaise about what I was writing. "Why the hell am I bothering writing this lot? It's a given that nobody's ever going to read it." Why? Because nobody reads documentation. Funnily enough, I tend to work the other way, too. While I will often go looking for help through Google, checking for assistance in structured, project-produced documentation (manuals, et al) is usually one of the last things I'll try. Why? Because nobody writes documentation. Now, given that I feel minimal motivation to write documentation because nobody reads it, and I don't usually read documentation because nobody writes it (probably because nobody reads it), an obvious question comes to mind: did one of these states come first, which then caused the other? Are we in an ever-deepening spiral of shite docs, the logical end of which is just giving up and reverting to code reading and experimentation?

Yesterday I landed in Subversion a branch of Tasks I've been working on to remove all confirmation dialogs and replace them with infinite undo/redo. I'm really pleased with the end result, there are no more dialogs getting in the way and every action is undoable. To implement this I adapted MarlinUndoManager from Marlin by the most excellent Iain Holmes. The basic design is that the application has a global undo context. When an operation which should be undoable occurs, you start the context, add as many undoables as required, and then end the context. Allowing a single undoable action to consist of multiple undoables lets the application reuse fine-grained logic to build coarse user-level actions (in Tasks this is used to build the Remove Completed action from multiple Delete Task actions). To make integrating this into applications even easier, I wrote a GtkAction which reflects the state of an undo context, so it is trivial to add redo/undo to the interface. There is one remaining task left before this is ready to be released: undo support in the edit dialog entries. Once this is done, the next release of Tasks will be announced. I also plan to work with Iain on cleaning up this code and submitting it for inclusion into GTK+. NP: Money Jungle, Duke Ellington, Charles Mingus, Max Roach

Working with subversion I’ve had a bit trouble making project wide search and replaces. find -name '*.pm' xargs perl 's/foo/bar/g' breaks a subversion working copy in not so subtle ways and grep’ing returns duplicate results. I gave up making the correct find statement in five minutes and installed ack. What a useful utility. Marc-Andr Lureau has ITP’ed it and it’s dependencies but hasn’t got anything uploaded yet. I offered to sponsor but havn’t recieved a respons yet and I’m impatient so I’ve rolled out my own packages, renaming ack to ack-grep to avoid name clashes with the existing ack package. Either download ack-grep_1.66-1_all.deb and libfile-next-perl_1.00-1_all.deb or use the following in you sources.list:

deb http://hacking.dk/debs sid main
deb-src http://hacking.dk/debs sid main

Still willing to either sponsor or maintaining the package myself.

Today was the day that Sarah was supposed to have an amniocentesis to determine if it was viable to deliver Joshua via caesarian section. Most probably he would have been born today. I should have become a father today. Instead, last night we stayed at a lovely Bed and Breakfast near Half Moon Bay, and this morning on the clifftops near the Ritz, we released a balloon with a photo of him, and cried a little. It's been a surreal couple of weeks. In some ways, it feels like it has dragged on forever, in other ways, it's hard to believe it has only been two weeks. I'm going to go back to work tomorrow. Sarah's healing from the surgery well, and her Mum's here, so there's no real need for me to be at home, and I think work will provide a welcome distraction. The support of all of our friends has been nothing short of amazing. We are truly blessed to have such a wonderful network of friends over here, which helps make up for the fact that our family is so far away.

On Friday I finally made the decision - from the two cameras that left in choice, I bought Pentax K10D. And first experiences are quite positive. In fact I didn't find anything really disappointing during first weekend of usage. I only learned, that I should better check manual choices I've made in the past, because changing white balance at interior, going out and taking photos there does not really produce great pictures. The camera feels really good in my hands and I was able to figure out most settings without reading manual, but I think now it is time to do so :-). Maybe it will help me to get even more out of it, especially in controlling manual modes. I have to admit that most of photos were left to automatics and I just tried what I can do with it. Combined with trips to Kyoto, Nagoya Castle and Inuyama, it is no wonder that I took almost 1000 photos during weekend. Thanks to everybody who helped me with their suggestions, it really helped me in the decision!

Stage 1 of the new koha installer uses MakeMaker and some silly tricks to try to install a CGI tree, as well as modules. You can often grab prepatches from mjr-installer if you really want the latest bits sooner. The biggest user-visible change with this installer (compared to 2.2) is that it's dumb. It's dumb in that it's trying to do the simplest thing that could possibly work and it's dumb in that it doesn't ask the installing user any questions. If you know you want to change things, you can tell it, but it doesn't ask you to confirm everything like the 2.2 installer did. Almost immediately, that was criticised on IRC. The first criticism was about a related feature that someone else removed earlier in 3.0, really. The second criticism was stalled by the realisation that if we really want an interface that "questions everything" every time, a wrapper script can ask for all the settings and instruct the dumb installer. I'm not sure whether asking users to confirm everything is a good move, for the sort of reasons in Havoc Pennington's 2002 essay under "The Question of Preferences" - each setting we encourage people to change makes it harder to debug installation problems. That's one of the biggest pains with 2.2 and I think we've started seeing the same happening in 3.0 already. It's not quite "no two installations are identical" but it's not far off. So, the main installer defaults everything to the best values that I can guess that will work within the constraints. It should work for most people and it should be easier for packagers to override. I'd still love to hear about how other Perl CGI scripts install, especially if they use MakeMaker for the basics, then use a CGI interface for the rest. Are you writing one?

For one UK bank, "innovation" has included funding long-term home loans mostly with short-term inter-bank loans. So, the increased costs of obtaining inter-bank loans during a credit crisis caused Northern Rock to obtain support from the central bank, the Bank of England. Once that became public knowledge, as it must, it started a predictable run on the bank, as described in Guardian Unlimited: Between Rock and a hard place - savers besiege bank. Savings are leaving at a rate of almost a billion pounds (1.6bn euro?) each day just now. In turn, that makes the bank's loans-to-savings ratio even more extreme, which then worries more savers and we're into a feedback loop, with queues growing out of the bank's doors, customer sit-ins [BBC West] and other unsettling side-effects. Northern Rock is a former member-owned building society which demutalised (converted to a shareholder-owned bank) in 1997. In just ten years, it has become committed to a business model which seems to scare many of its savers. Would it have done that if it was still controlled by its savers and borrowers? Is this exodus of savings going to be matched by loan customers moving away too? Will any of these customers return to the mutual and co-operative sector? It now seems sort of academic whether Northern Rock is on the list of online banks that work with free software, as commentators suggest it may be taken over soon. 2007-09-18 Update: the UK government has guaranteed all deposits in the dodgy bank. This is a bad thing, as explained on niq's soapbox: Race to the bottom.

As uclibc's removal from Debian has been requested already, it is really time to push forward with this package. The technical bits of the package haven't changed in ages, and could be said to be pretty solid (I use the package daily). The problem with this package in the Debian context is that the licencing is somewhat less than clear, as no one really cared about adding their name to the list of copyright holders. I can see a few different variants how to deal with this: Number one would be the "massive audit" approach. This is the first thing I'm going to try, simply hang out on IRC for a few days and try to find people who join me (on freenode's #uclibc, mostly). If that doesn't work out, there will be lobotomized uploads, where I start a "new" package and subsequently move the functionality back in, in whatever tempo I can manage, until I get back to "pristine upstream source".

Although slow, development on svn-buildpackage still continues. Sorry for the delays, but both myself and Eduard Bloch have been unable to assign more time to svn-buildpackage development in the last few months.

From the changelog of the trunk package (unreleased 0.6.22):

  [ Eddy Petri or ]
* IMPORTANT: changed default behaviour of saving the configuration in
.svn/deb-layout by default to avoid stale data to override the
configuration options that were updated in the repository.
(Closes: #414581)
As a consequence, a new option --svn-savecfg was added to allow a
mechanism for easily overriding options locally

BTW, did you know that svn-buildpackage is maintained in collab-maint and welcomes contributions and committers/reviewers (especially those)?

In case you want to help, this is the place to start from:

svn co svn+ssh://svn.debian.org/svn/collab-maint/deb-maint/svn-buildpackage/trunk

or

svn co svn://svn.debian.org/svn/collab-maint/deb-maint/svn-buildpackage/trunk

If you don't have commit access to collab-maint yet (just make a request on the alioth tracker and that will be fixed ;-) ).

I have just read an interesting post by Ted Ts’O about copyright protection on the net [1]. Ted is well known as a free software programmer, but it’s slightly less well known that he is an avid Science-Fiction fan. In the Free Software community most people seem to be interested in Sci-Fi, but Ted is more interested than most. Ted’s post concerns the irresponsible actions of the Science Fiction and Fantasy Writers of America [2] (SFWA). To summarise it they issued DMCA [3] take-down notices for any web page that matched a search on the names “Asimov” or “Silverberg“. I don’t approve of the DMCA laws as a collection, but take-down notices are not necessarily bad (I have issued such notices for unauthorised copies of my own work in the past). The problem in this case is that the words in question are extremely common, not only might they be used as the names for other people (authors or characters) but Asimov in particular is a well known term when describing the potential development of intelligent computers that operate robots (see the Three Laws of Robotics [4]), the term “not Asimov Compliant” has been used by Alastair Reynolds in Century Rain to refer to a class of military robots that have no compunction about killing humans. Among the fall-out of the SFWA actions was the removal of a free novel by Cory Doctorow [4]. Incidentally my favourite free to download Cory Doctorow book is Eastern Standard Tribe [5]. craphound.com has Cory’s blog as well as links to other free Sci-Fi that he’s written. Reading the links from Ted’s post took me to a blog entry by the current SFWA vice-president [6] which describes authors such as Cory Doctorow as “webscabs“. This offends me greatly. My work and that of my friends in writing free software could be described in the same way (and in fact is described in a similar way by some software monopolists). Every blogger could have their work described in a similar way by paid journalists. The fact that the SFWA VP is not representing SFWA when writing such comments does little to allay concern about this. It seems to me that people with such ideas are intent on attacking my community, and that it would be wrong of me to give any of them $0.50 by buying one of their books. I resolve to not buy any more Sci-Fi books until I have read all the freely available books that I want to read. After that I will prioritise my book purchases with a significant factor being how well the author gets the concepts of copyright etc. If nothing else an author who can’t understand how copyright (something that is essential to their own livelyhood) interacts with current computer systems will have significant difficulties in predicting how technology and society will develop over the next hundred or thousand years. My problem in reading Sci-Fi books is not in discovering books that are enjoyable and which contain interesting concepts, but in finding time to read them. Thanks to SFWA for giving me an extra criteria to cull the list of books to read.

• [1] http://thunk.org/tytso/blog/2007/09/15/whack-the-gopher/


• [2] http://www.sfwa.org/


• [3] http://en.wikipedia.org/wiki/Dmca


• [4] http://www.boingboing.net/2007/08/30/science-fiction-writ-1.html


• [5] http://craphound.com/est/


• [6] http://community.livejournal.com/sfwa/10039.html

Share This

When packaging games, one of the operations I often have to do to the source code is modifying it so that it uses absolute file and directory names instead of relative ones. When coders develop for Windows, or when they develop their games to be able to play them from their home directory, they just would go that way and won’t care about it, but when the idea is installing the files in the system, the Filesystem Hierarchy Standard (FHS) must be respected, and thus things will need to be place in their proper directories. For read-only files, this is not a big deal, all that has to be done is replacing the relative file name in the file opening commands with the absolute one, doing fopen(”/usr/share/games/package/dir/file”) instead of fopen(”dir/file”). I usually prefer to do it like fopen(DATADIR “/dir/file”) and set DATADIR from the Makefile, with a -DDATADIR=\”/usr/share/games/package\” switch added to the CFLAGS, and maybe also setting it inside the code if it’s not previously defined (#ifndef and so). There are some files and directories in which doing this is not possible, because they want read-write permissions. This files include saving the configuration, hiscores, new levels created with the game editor, game saves and so. They might be moved to some directory under “/var/lib/”, and create a new group en the system so that it’s writable, and so on. I don’t like that approach a single bit. I prefer to move all tthat data to the user’s directory. This is not as straightforward as it was for read-only data, but it’s not really that difficult anyway. It would be more or less like:

#ifndef _WIN32
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>
#include <limits.h>
#include <pwd.h>
#include <fcntl.h>
#include <sys/stat.h>
#include <sys/types.h>
#endif
#ifndef _WIN32
char file[PATH_MAX];
char *home;
struct passwd *passwd;
if (!getuid()   !(home = getenv("HOME"))) 
passwd = getpwuid (getuid());
home=passwd->pw_dir;
if (!home) 
fprintf(stderr, "$HOME is not defined.n");
exit(-1)
 
 
if (strlen(home) > PATH_MAX - sizeof("/.package/settings.ini")) 
fprintf(stderr, "$HOME is excessively long.n");
exit(-1)
 
snprintf(file, sizeof(file), "%s/.package", home);
mkdir(file, S_IRWXU   S_IRWXG   S_IROTH   S_IXOTH) == 0;
strncat(file, "/settings.ini", sizeof(file)-1);
#else
char file[21];
snprintf(file, sizeof(file), "./Data/settings.ini");
#endif
FILE *f = fopen(file, "w");

Sometimes you also need that the files are already created, you can check if they already exist with stat(), and then acting accordingly, either creating them from your C code, or maybe triggering a system() shell call whith whatever is needed.

struct stat stat_buf;
stat(file, &stat_buf);
if (stat(file, &stat_buf) == -1) 
int settings_fd = creat(file, O_WRONLY);
close(settings_fd);
 

NOTE: I’ve updated the code to add some suggestions. Thanks to Gon ri Le Bouder and Nico Golde.

Brice already mentioned it, but I've uploaded Xorg 7.3 to unstable this afternoon. This includes Xserver version 1.4, which has improvements all over the place, perhaps most notably by including input hotplugging (via HAL) and many improvements in EXA, the next generation 2D driver acceleration architecture.

Now that this is uploaded, I'll be ripping out large parts of the xorg.conf generation script. All the various improvements over the past few releases have made it possible to do this, and for lenny, if all goes according to plan, we won't have to generate a xorg.conf for installs at all. You can still use one (and many people will want one to customize their settings in some way) but for the most part people shouldn't have to bother with them. I'll be spending a lot of my time over the next few months tightening this stuff up so that a lot of the little things really work the way they should, and now that we have the updated server available I'll be able to actively work on these things in Debian so you'll be able to see the results quickly.

On a side note, I went to Software Freedom Day yesterday. I heard ari give a talk on the Gimp. At the last minute I almost gave a lightning talk on where X.org is headed, but they ran out of time. It would have been kind of weird though, since the talks were on things like software patents, free culture, and miro (formerly democracy-player) rather than the nuts and bolts sorts of things that I'm more comfortable with.

FYI, my new Miro packages (formerly known as Democracy Player) have now reached unstable. After lots of ugly, ugly trouble with even getting a successful build (boost/python/dbus related, you don't want to know) the packages are back in shape now, with tons of fixed (or no longer reproducible) bugs and lots of upstream impovements and new features. If you reported a bug against Democracy Player, please try the latest Miro package and check if it still occurs, thanks! The upgrade should be seamless, your existing config and videos will be migrated from ~/.democracyplayer to ~/.miro automatically upon the first start of Miro. Some of the new/fixed things in this release include:

• HTTP proxy support (uses the GNOME proxy settings, use gconf-editor to change them).
• Flash videos now play fine (non-jerky) and with sound!
• You can search various video sites (Youtube, Google video, etc.) online, and even save searches as channels.
• You can export your channel list into an OPML file (and also import OPML files, of course). I've been waiting for this for a very long time (it's a good way to backup your channel list, or move it to another machine)...
• Lots and lots of bugfixes and small enhancements, as usual.