When looking at their screenshots of their to-come office 12 user interface I found some things that are just dead wrong.

Basically they help users to choose a pure visual instead of information oriented approach to documents. The semantic formatting tools are discredited as "quick styles", totally ignoring their behaviour (for me this implies that this is just a set of formattings, not e.g. used for a table of contents, and also not that the formatting will change when I change the style definition) In the age of XML this is just plain wrong!

Look at one of the first two screenshots. The semantic formatting is moved to the right, becoming like a second-class formatting, with the first class being the font face, and even the underline formatting tool...

(Underline is considered to be an outdated formatting, back from when your typewriter didn't have any formatting options, and underline was the only thing you could easily add manually... it is bad, and you shouldn't use it except for hyperlinks!)

In a modern tool, the font box there should not exist. All formatting should be done by assigning semantic "tags" and defining how these are expected to render. (Just like you have been doing with LaTeX for ages, and many people are doing nowadays with CSS.)

And the "font" dropdown especially is really bad. Nothing is worse than letting people use different fonts within a document that easily. Hide all font choosing options except for letting them pick one font for headlines and one for the text body. The results will be way better.

Basically, Word is still not a "document processor", but a paint program. The tools are more suited for making a poster than for writing a document...

A good word processor would help people separate formatting from content, instead of doing the wrong thing easier.

Also, from a very different UI level: the "tabs" don't look like tabs. But I guess that will change with the final skin.

Still it looks to me very much like just a revamp of the UI, and maybe some nice gimmicks such as converting lists to diagrams, but it is lacking the real step from the visual "document painting" to a real next generation document processing.

Oh, and the Microsoft "XML format" will be pretty useless for exactly this reason: it doesn't help users to markup the semantic meaning, so the word XML files will be mostly "enable bold" "enable cursive" etc. - useless.

I just got flashblock installed on galeon (which is still my favourite browser, not firefox). Epiphany probably works the same way.

Flashblock will replace all flash and java object elements with a button you can click to launch the contents. That is very convenient (no more annoying animated-emoticons flash banner ads eating your CPU) - and increases security (less risk of being infected via a security hole in a plugin).

So here is what I think is sufficient to install it:

1. Go to "about:config" and set "xpinstall.enabled" to "true"
2. Install flashblock (I had to download the .xpi file and install it from the filesystem)
3. If you already had a userContent.css, add to the beginning the following statement: @import url(chrome://flashblock/content/flashblock.css);

I love cooking, especially with good friends. And I love food. And I've got dozens of food pictures, since we started doing these during my stay at Berkeley... a nice way of saying "we're fine" is to show off the food you are doing. Life is doing well if you can enjoy making your own good, healthy food... This shot is of my latest Chili con carne dish. Theres is rather little meat in it, but lots of tomatoes, bell pepper, corn and beans. I also eat lots of rice with it.

Ich liebe es zu kochen, insbesondere mit Freunden. Und ich liebe essen! Inzwischen habe ich einen Haufen Fotos von Essen, da wir in Berkeley irgendwann angefangen haben unsere Kreationen zu fotografieren... es ist irgendwie eine nette Art zu zeigen, dass es einem gutgeht, indem man das leckere und gesunde Essen das man sich eben gekocht hat herzeigt. Das Leben läuft gut, wenn man in Ruhe und genüßlich sein Essen kochen kann... Dieses Foto zeigt mein Abendessen von gestern und für morgen: Chili con Carne, selbstgewürzt. Wenig Fleisch, dafür um so mehr Tomaten, Paprika, Mais und Bohnen. Dazu kommt noch ein Topf voll Reis...

Mmmmmh! Food!

I just read in a ticker about someone from italy discovering an ancient villa by staring closely enough at google earth satellite images...

So I entered the location they mentioned into google maps: "Sorbolo, Italy" switched to satellite mode and zoomed out until the image quality was available.

Dear, google, this is not Italy. Not at all. This is about as far off as you can. That is called the "south pole" usually, not Italy.

My Layered Subversion script now can query the user for authentication credentials.

That way I've deployed the first machine with laysvn. ;-)

Next I'll really have to do the add and commit stuff, so I can easier get my changes back into the version control. (well, copying the new files into their shadow place, then calling svn add and svn commit there is not bad)

If you are using laysvn, please send me an email, I'd like to know if anyone is actually playing around with it, or if just everybody says "cool, exactly what I need" but otherwise waits for me to finish it... ;-)

I've had a hard time explaining on the mercurial list what I want to achieve. I think they still want me to use merge for that. Which is annoying... Again there is no point in talking to anyone working on a SCM when you've mentioned another SCM... they will always react like you insulted their project just by having worked with a different SCM before.

The results in todays election in germany are very surprising. Polls had shown a close call for the opposition parties (CDU, CSU and FDP) to take over government. But the first results shortly after the booths had closed gave a very different view, and this didn't change much during the evening (the counting is still in progress).

The "to-be winner", the conservative CDU-CSU sister parties (sometimes named "Union"), are around 35%, which is 3% short of their previous result, likely to become their third worst result ever and is 6% short of their result in the polls. The previously biggest party, the german socialists, dropped to 34%, which had come out at the polls. The partner of the Union partys, the german liberals (don't confuse this with "liberals" in the US, this is a lot about having less regulations... but they do civil rights, too) had a surprisingly good result with 10% (they were expected to have 6.5%). The german greens, which are also part of the current government, lost slightly to 8%, also one percent above the poll results. The left-wing "Die.Linke" party, a joint party of the PDS party - said to be the successor of the communist SED party of former eastern germany - an a spin-off of the "socialists" SPD, achieved the expected 8-9%.

This brings an awkward situation to germany: the expected coalition of CDU-CSU and FDP does not have enough mandates to form government, the current government is even further off. Noone is willing to cooperate with the left-wing (which in my opinion has no sense for reality and no working concepts, and the only good thing they achieved is to make the nazi right-wing partys lose any significant share...).

This leaves basically three options: socialists and conservatives, dubbed the "grand coalition", the "traffic lights" combination of socialists, greens and liberals, and the "black traffic light" or "jamaica" coalition of conservatives, greens and liberals. I would favour the latter, since this would mean an actual change.

To make it even more weird, the liberals have announced that they will not form government with either the socialists or the greens (since they promised to do away with the red-green government). Including the conservative-led jamaica coalition, where the greens would be the weakest partner of all.

So we have only the conservative-socialist coalition left (which in turn would make the liberals fail, since this means red is part of the government because of them!) - which will hardly work out. The conservatives have done a really bad campaign, which basically consisted of "7 years red-green is enough", and other "socialists are stupid" slogans...

Now both leaders of the conservatives and of the socialists claim to be the winners of the election (both parties have an historically bad result) and want to become chancellor...

However this will turn out (my bet is that the liberals change their mind, as usual...) - it will be interesting to watch.

Because of this situation, it is next to impossible to tell what the outcome will mean for open source software, civil rights or bicyclists. ;-)

This obviously is an american citibank branch. In San Francisco, Chinatown. At first you smile, but when you see the banks one by one each looks more fake than the others. They do not blend in, however hard they try... And US banks are really bad, compared with german banks. Well, as is banking in the US in general. And don't get me started on your identity theft being your own fault.

Das ist offensichtlich eine Citibank in den USA. Genauer gesagt in Kalifornien, San Francisco: im Chinatown. Auf den ersten Blick ganz "putzig", aber mit jeder Bank wirds schlimmer, eine schaut dann falscher aus als die andere. "Es ist nicht alles Gold was glänzt". Banken in den USA sind sowieso eine Katastrophe, verglichen mit den Banken in Deutschland. Und überhaupt: Schecks, Schulden und Social Security Numbers als geheimstes Geheimnis (man braucht sie nur überall, aber wenn sie in die falschen Hände gerät heisst das "Identity Theft") - das charakterisiert den Zustand in den USA ganz gut.

Chinatown Citibank

You can now diff files within the "layered" checkout directory. It won't work for directories though. Similar (since it's pretty much the same code) restrictions apply to the new revert command. It will also only work on regular files.

I'm also not sure how these commands will work on files that were removed. They might work just fine, I havn't tested it yet.

Anyway, I can already use this tool to verify the configuration of a server of mine versus my reference config. And by checking in my storage dirs I can also commit changes. ;-)

You can download it as usual from my LaySVN page.

The latest mentioning of mercurial on LWN got me interested in that one though. It's written in python, too.

Another shot from my US trip: this was at Bridgeport reservoir. Since tioga pass of Yosemite was still closed we had to take sonora pass, so we arrived at bridgeport reservoir in the late afternoon. Just in time to get to a campground, prepare dinner and enjoy the sunset. That was also the coldest night in the US we had.

Nochmals ein Foto von meiner USA-Reise: diesmal beim Bridgeport reservoir. Der Tioga-Pass aus dem Yosemite war noch geschlossen, deswegen haben wir den Sonora-Pass nördlich davon genommen. Bei Bridgeport waren wir genau passend für das Abendessen und zum Suchen nach einem Campground. Der Sonnenuntergang war sehr malerisch, der Campground hatte das schönste Bad von allen - die Nacht war dafür sehr kalt, so dass wir die Heizung anmachen mussten.

Dusk sky at bridgeport reservoir / Wolken bei Sonnenuntergang

(yesterday I did not have time to update my photoblog...)

A rather old meme was just revived...

create your own visited countries map

Looks like much because of US and Canada, but I really can't keep up with others... maybe after my diploma... my brother did travel around a lot in asia, that would be very interesting.

This is an "old" shot, from June. These are the sand dunes in Death Valley, at sunset. What I particularly like about this picture is the odd colors. This is a result of deliberatly enabling the flash to illuminate the foreground. I guess a "pro" with a good film could have produced nicer colors for the sky though, and maybe a bit more contrast on the foreground.

Anyway, for a spontaneous shot that took just a few seconds I'm pretty happy with it... ;-)

Das ist ein "altes" Foto, aufgenommen im Juni im Death Valley. Dort gibt es Sanddünen, die im Sonnenuntergang besonders toll sein sollen. Also sind wir nach der Hitze des Tages dorthin aufgebrochen. Aus einer spontanen Idee heraus hab ich dieses Foto so gemacht, und ich bin überrascht wie interessant es geworden ist: Die eigenartigen Farben kommen zum Teil durch den Blitz, mit dem ich absichtlich den Fordergrund aufgehellt habe.

Ein Profi mit einer guten Spiegelreflex hätte da sicher mehr rausgeholt (insbesondere beim Himmel), aber für meinen kleinen flinken Knipser (der m.E. einen schwachen Blitz hat) und die spontane Situation bin ich mit dem Ergebnis mehr als zufrieden.

Dusk dunes / Dünen in der Dämmerung

LaySVN has now svn:ignore support. Well, kind of. I read that property and have my own global ingore list, then use the python fnmatch module to find out which files I should ignore. I didn't find a way to ask subversion if it would ignore the file.

Some stuff is not working right, and I still havn't worked on any "commit" type of stuff. But you are welcome to extend it... I was kind of disappointed by not finding in the pysvn and svncpp docs what I was looking for. :-(

Grab the latest version at the usual place.

Maybe a different SCM is better suited, but I'm not sure which ones offer any language bindings at all?

James Cape blogged about Organic HTML.

An interesting idea, visualizing a web page structure as a flower. ;-)

Unfortunately, typical CSS pages look pretty boring that way. It's the "evil" HTML elements that make a fancy flower.

msn.com makes the flash applet crash my browser at first try... planet.debian.org looks like a jungle, not like a plant...

Not much today, just a shot of the brugmansia tree on our patio. When the sun came out I wanted to try out manually controlling the aperture to get the background blurred.

Heute nichts besonderes, nur ein Foto von der Engelstrompete auf unserer Terrasse. Ich brauchte was zum spielen als die Sonne rauskam, und so hab ich mal manuell die Blende eingestellt um den Hintergrund unscharf zu bekommen.

Burgmansia flower / Engelstrompete

I'm stuck: I can't get subversion to tell me if a file is to be ignored in a repository. That is a file which is not yet existant in the working directory...

I first thought this might be a limitation of the pySVN api (which e.g. doesn't expose the singleStatus of the C++ API), but the C++ API also returns a None-Status. So I started browsing the subversion source... There is a function called svn_wc_get_ignores in there which internally collects the ignore patterns, but I can't find an exposed version of it I could use.

So I see two choices: going away from subversion to some other SCM as base, or restricting the support of ignores to those defined in svn:ignore and a set of builtins of laysvn (like *~)

I can't understand all the hype around Ajax (Asynchronous Javascript and XML).

I've done some experiments with it, and there are so many caveats...

Basically, most of the time you are hacking around bugs and missing stuff in browsers. If you look at google maps, this is between 150k-200k in JavaScript source, optimized javascript that is with two-letter function names. For each browser supported, totaling in over 500k. In freaking compressed javascript!

That is not the future. That is stone age!

Now to OpenLaszlo, which was just mentioned on Planet XMLHack.

Don't even bother to look at it. That is even worse! It's using Flash, which is a usability crime. Is there any free parser for flash? Is there any (working with all recent extensions) other application than the flash plugin itself that can playback flash?

Even worse: is there any screenreader that can access flash? A search engine which can get the actual information from flash?

No. Flash is the way to go if you want to make your website really really inaccessible. Don't do it. With recent CSS we have reached a point where you can write XHTML pages that contain the actual information and put all the fancy layout stuff into CSS so screen readers and search engines don't get confused. So that third parties can access your site just fine.

With XSLT and XML this can become even better. That is what the semantic web (or whatever marketing people refer to by "Web 2.0") is about, not some fancy graphics stuff: making information accessible.

And maybe in the far far future, Microsoft Internet Exploiter will maybe be able to support all recent standards correctly. Like CSS. And maybe SVG, so we can get rid of Flash completely.

The second aspect with "Web 2.0" that I'm aware of is Human-Human interaction, btw. This has changed a lot in the past few years, independently of hypes such as Ajax. By reading this, you basically acknowledge this: Blogs have become something quite important in "todays" web. And they are just one part of the puzzle. There are friend-of-a-firend networks, podcasts, and while this is also "fancy new stuff" and might even use Ajax, the point is about information interchange between humans. Write your blog in Flash and it can't be syndicated. Make your blog in ajax and google won't index it except if you write a special "old-style" Google backend, too.

I just released laysvn 0.1 (if you wonder what layered subversion is, visit that page, or read this blog posting of mine).

This "release" can only do a checkout/update for the layered repository. But by running it and then changing to your storage directories (this is where the real subversion checkouts reside) and running commit there you can already commit, too! Jippee!

The empty platform of a train station outside of munich illuminated at night, with the camera placed on the floor for an unusual perspecitve. Another shot between reality and a computer graphics like feeling (have you ever played first person shooters? the texture of the floor looks blurred like in typical computer games at these angles, and the train stations structure is so regular it could have been a computer construct...)

Der leere Bahnsteig einer S-Bahn-Station ausserhlab von München. Eine Insel des Lichts in der Nacht... Die Kamera ist auf den Bahnsteig gestellt, was für eine eigenartige Perspektive sorgt, irgendwo zwischen realität und Computergrafik: Die unschärfe der Boden-Textur im Vordergrund erinnert an Computer-Texturen, und die Träger des Bahnhof-Dachs sind so gleichmäßig konstruiert, dass sie leicht aus "virtueller realität" stammen könnten.

Take me home...

I've been hacking code todays quite efficiently... my "layered subversion" client is coming along nicely.

You'll probably wonder what "layered subversion" is supposed to be.

Basically it's an overlay of multiple subversion repositories, the same thing you would get by exporting them and then copying them over each other.

Except that I want to be able to check-in my changes to the repository where the file came from...

So I e.g. have a configuration like this:

Configuration 'Host foo /etc' has the following layers:
Layer ID: 'base' is 'http://confighost/svn/layers/base'
Layer ID: 'selinux' is 'http://confighost/svn/layers/selinux'
Layer ID: 'firewall' is 'http://confighost/svn/layers/firewall'
Layer ID: 'foo' is 'http://confighost/svn/layers/foo'

I have a set of configuration files common for all my servers. These files are stored in the base layer, and when I update my machines I want to recieve the latest updates.

Then I have the SELinux layer, which e.g. contains policies, but also overrides some of the default init scripts (removing stuff which won't work on SELinux due to security restrictions, e.g. updatedb)

The third layer is shared for my firewall boxes, whereas the fourth layer contains e.g. /etc/hostname which is unique for this host.

Now assume I've changed some files in my configuration (e.g. by runing apt-get upgrade), I now want to commit these changes to their respective layers. While I could do this manually, or by merging somehow etc. I want this to be automated as far as possible. In most cases, files won't move from one layer to another; even the case where a file exists in multiple layers (the SELinux cronjob example above) is very rare.

Here's an output of my current "laysvn status" command:

M foo   fstab
? ---   foobargnarf
M base  magic
? ---   foo

There is still lots of stuff missing. Some of the code is also rather ugly, it's more of a hack into the python subversion API: the only way I've found to query whether a (not yet existant) file is to be ignored (in which case I will not look in lower layers, so I can "delete" files in upper layers!) or if it is just a new file is really odd in pysvn: a new file will give an empty result for the status() call, whereas an ignored file actually is found as ignored... it would be nice to get a proper "404" return code in the first case.

I have no idea yet of the semantics to be used in conflicts, or which issues will arise with the "update" call. I have a rather clear view of the latter through...

When I'm at a point where I can use this tool for my needs I'll release it, and hope that someone else picks it up and adds the functionality to either regular subversion or svk... ;-) Long live OpenSource!

Many of us developers will remember when XFree86 changed it's licence with the release of 4.3.something (officialy with 4.4). When Debian decided to stay with the last version that still had the old licence and backport drivers selectively that do not have the new licence terms (which were most of them IIRC. I remember Alan Cox explicitely not agreeing to have his code's licence changed)

Noone ever noticed Xfree86.org releasing 4.5, did you?

Xfree86.org has gone entirely off the FLOSS radar.

In retrospect, the project killed itself by changing their licence, putting in some clauses many users, developers or maintainers found inacceptable. (FSF says the licence is incompatible with the GPL, i.e. a GPL program cannot use the xlibs from xfree86.org)

Everybody is using X.org now. (well, we at Debian are still shipping xfree86 only in sarge, but we are only providing security updates for it anyway, and we have an updated version to which the old licence still applies).

And I actually think that it was a good thing:

X development has picked up a lot since back then, I have the impression that more people are actually working on it than before. (But that may also be due to the joint efforts of GNOME and KDE at freedesktop.org)

With X.org we've seen a couple of new developments that have long been requested for Linux, including the famous Screenshots with real transparency and real shadows by Keith Packard which made users scream for X.org at the poor Debian maintainers... (although performance depends very much on your driver, and maybe stability and other archs are more important...)

One of the bigger things for developers - which will happen with the 7.0 according to schedule - is the modularization and autotoolization of X.org sourcecode, which should make maintaining them a lot easier apparently. Yay!

Another shot from friday night: a childrens' playground scarely illuminated by the street lights. Actually it is a very new playground, and there is only a walkway here, no road.

Noch ein Foto von freitag abend: ein Kinderspielplatz, der von den Straßenlampen nur spärlich beleuchtet wird. Es ist sogar ein sehr neuer Spielplatz, und keine Straße in der Nähe, nur der Fahrradweg an der S-Bahn.

Noone is playing here / Niemand spielt mehr hier

Nachdem meine Kamera endlich von der reparatur zurück ist, habe ich jetzt mal ein Photoblog angefangen... keine Ahnung wie regelmäßig ich das aktualisieren werde...

Da meine "Syndications" (Planet Debian, Planet XMLhack, Munichblogs...) alle entweder den deutschen oder den englischen Teil betreffen, aber die Fotos in beides passen, hab ich sie zu einer eigenen Kategorie gemacht.

Naja, und wie ihr sie in euren RSS-Reader aufnehmt etc. wisst ihr eh selbst. Wenn ihr mich gerne in einem "Planet" hättet, schreibt mir doch ne Mail.

Hier ist noch mein "Eröffnungsphoto":

Since my camera just came back from being repaired, I just started a photo blog. Most of my syndications (Planet Debian, Planet XMLhack, Munichblogs) are either german or english only, my photoblog is somewhat independent, so I made the photoblog separate. If you are interested you know how to add it anyway. If you'd like to add it to a planet, just drop me an email.

I don't know if I'll be updating it frequently... anyway, here's my opening photo (titled "the last train") to get you interested (or not...):

I open my photoblog with an experimental shot I did last night:

Ich eröffne mein Photoblog mit einem experimentellen Photo von letzter Nacht:

The last train to munich rushing by.

Der letzte Zug nach München rauscht vorbei.

I love riding my bicycle in Munich. Once you know the hidden roads with low traffic or where there is a nice bike lane it's really fun.

I especially like riding back my bike at night (at least when it's not too cold, some rain is okay, since I'm going to take a shower afterwards anyway) - it's so peaceful and quiet. I've seen a fox, I've seen rabbits. I often scare cats...

I'm pretty fast with my bike - usually faster than with public transit, and searching for free parking in Munich sucks a lot. Also you are much more free to do what you want: drink some beers - doesn't matter much (I'm not driving much on the roads...) - stay as long as you like (riding a bike will keep you awake, and you also do not have to catch the last train home or pay lots for a cab).

My typical distance is 12-16 km. I need 45min to one hour for that, and this is a distance I can do at "full" speed without being tired afterwards. I've done much longer trips of course, but you ride these differently.

Today I did a couple of photos, to make a web page explaining my route, to encourage others to do the same. By doing that (and by seeing a couple of city tour busses) I noticed how many nice spots in Munich I come by everyday - my route is almost a city tour! And with the bike you can just stop, enjoy the scenery, take a picture.

Whenever I have some time in a city, I try to walk around a bit. Without that you'll never really know how the city feels like, how life is like there. In San Francisco (well, I'm not an insider, but I think many people have been to SF, so some might know what I'm talking about), I recommend you to walk from mission district to Castro, and maybe a second day from Golden Gate park to the bridge, then maybe on to Fishermans Wharf. A third tour would be from Embacadero to Telegraph Hill then Chinatown, Columbus Ave and to Fishermans Wharf. Maybe take a cable car back then. You'll see very different sides of the city this way. The tourist side, the shopping parts - and the quiet living quarters in Richmond.

Munich is much safer than any US city. It's in fact one of the safest citys in the world. So you can walk around there at night without feeling threatened (except by crazy car drivers). Since I'm mostly using my bike there I can't really recommend routes for walking, since I don't have a good feeling for walking distances. ;-)

So here are some recommendations: Walk from Hauptbahnhof to Marienplatz. This is the heart of Munich, a big shopping zone and Marienplatz is very touristic. Now turn north, and you'll pass by the Feldhernhalle, which has an historic role in Nazi Germany, btw. and you'll end up at the Ludwigsstraße. This is maybe the biggest street in Munich, and there are lots of "classic" Munich buildings there. A couple of official stuff, but also the university. The place at the university is called "Geschwister Scholl-Platz" on the left side, and "Professor Huber-Platz" on the right, to honor their resistance against Nazi Germany. Next you'll see the Siegestor, which was built in the 1840s to honor the bavarian military. Since world war II it bears the words "Dem Sieg geweiht, im Krieg zerstört, zum Frieden mahnend" (dedicated to victory, destroyed in war, reminding to keep peace") then continue on the road (now called Leopoldstraße) through a quarter with lots of bars and cafes. When you reach "Münchner Freiheit" which is a very nice place in Munich, get some ice cream, then turn right and walk random streets until you reach the "Englischer Garten". This is a huge park in Munich, and where many people relax. Walk south-east, following the Eisbach (a small stream) until you've seen the surfers (yes, you can do surfing in munich, on this small stream!). Then turn left of Prinzregentenstraße. You'll come by the bavarian national museum. When you reach the Isar, you'll see the "Friedensengel" (Angel of peace) statue on the other side. Go there, then continue south on the east side of the river (another part of the "Englischer Garten" stretches there). At the next bridge is the bavarian "Landtag", this is where the parliament of bavaria meets. At the south-east end of the bridge, find your way uphills through a park. Behind a small childrens' playground you'll find a "secret" entrance to a "Biergarten". You should be able to hear it from a distance - whenever there is good weather there are people there chatting and enjoying their beer and food. This already makes a nice end for the trip, but you can also continue to the east where there a dozens of nice cafés and "Kneipen" up to the "Ostbahnhof", which is the second most important train station in Munich. On the east side of that you'll find dozens of clubs to spend the night in - this is the most famous party zone.

I can't use my gmail account currently. Something is broken. Others don't have the same problem, so apparently I'm on a "bad server"...

Server Error

The server encountered a temporary error and could not complete your request.

Please try again in 30 seconds.

Well, I've been trying a couple of times the last 4 hours or so... The farthest I've gotten is to the "Loading..." display.

Update: I can now login again. And emails start tickling in that have apparently been on hold for some 6 hours.

The Winner of the Gnome 2.12 splash contest is out. Interesting picture, but to me it doesn't look "evolving", but "dissolving".

I think of detergent tabs dissolving in water when I see the splash.

In two weeks there are elections in germany for the state parliament.

I'm a member of the green party in Germany, and they again reflect my views very much. They focus a lot on privacy, freedom, equality and peace.

It is very unlikely that they will be part of the upcoming government (but they have been for the last seven years; unlike to the US where it's basically an either-or voting, we have more realistic choice here). It's 99% sure that the conservatives will be leading government, either with the german socialists (who have nothing to do with socialism) or the liberals. Their campaign is focussed around unemployment, and basically they promise that unemployment will decrease by cutting social security. They also want to increase security a lot (not that we have a high terror risk anyway...) and don't really care about privacy, freedom, equality or peace. They would sacrifice any of them to increase economical growth (because they expect that this would decrease unemployment, although there is serious doubt about that... it's not like we wouldn't produce enough. We produce more than we can sell, basically...)

Anyway, it's been very disappointing watching their campaigns. People don't care about their privacy, they want somebody to tell them "we're going to do everything different (but don't yet know what exactly), and it will be better afterwards". Or - the new left-wing party - "we're going to do everything dfferent, and it will be better afterwards (when we find out how it actually works)". While a party like the greens, who have actively been pushing e.g. regenerative energies (it is expected that in a few years wind power will be the cheapest energy source) - and germany is very strong at that, and there have been thousands of new jobs in this field - is not being credited for that. People want promises, not real solutions. :-(

Gestern abend war das TV-Duell. Da unser lokaler Elektromeister es nicht geschafft hat, innerhalb von einem Monat uns ein Angebot über DVB-T und DVB-S zu machen (unser Fernseher steht im Keller, aber wir sind eh nicht mehr in dem Bereich, wo eine "Zimmerantenne" reichen würde), haben wir das Fernsehduell bei Verwandten angeschaut. Ansonsten sind wir derzeit ohne TV, was auch kein großer Verlust ist.

Zum TV-Duell muss ich sagen, dass mich Schröder mit seinen Pointen gegen Merkel ("wir reden hier über Bruttolohn, ja?") das ganze wenigstens witzig gemacht hat. Oftmals steckt da auch viel wahres drin ("Mehr Schuhputzer für Computerspezialisten" - Arbeitsplätze kann man nicht herzaubern, sondern es fehlt bei uns immernoch an der Ausbildung. Manche Leute können halt nur "Niedriglohnjobs" machen, wenn man nicht Milliarden, die auch wir nicht mehr haben, in Fortbildungsmaßnahmen steckt. Das besagen auch aktuelle Umfragen, und daran kann die Union auch nichts ändern. Merkel würde es nicht schaden, mal eine längere Zeit in den USA zu verbringen, dann wüsste sie wirklich, was echte Niedriglohn-jobs sind... Maut an Brücken kassieren, den ganzen Tag im Autodreck und den Abgasen...)

Merkel wirkte ihm gegenüber etwas hilflos, insbesondere da ihre Partei ja eben genau keine Maßnahmen nennen kann, mit denen sie die Binnennachfrage (wir sind ja inzwischen wieder Exportweltmeister) ankurbeln könnten, aber gleichzeitig die Mehrwertsteuer erhöhen. Wenn man ausser der "Wachstumslüge" nichts zu bieten hat, ist es natürlich schwer gegen den "Medienkanzler" anzukommen...

Naja, meine Meinung habe ich ja schon zuvor dagelegt. Ich wähle Grün, weil ich eine starke Opposition will. Die Regierung wird entweder Schwarz-Gelb oder Schwarz-Rot werden. Grün wird sicher in der Opposition sein - und das können sie eh von allen Parteien am besten - und da wir eine starke Opposition brauchen, brauchen wir viel Grün (und auf jeden Fall mehr als die prognostizierten 7%. 10% Grün wären toll, sind aber illusorisch!)

Die Union hat eigentlich alle Themen vernachlässigt ausser "Arbeit". Die Themen Privatsphäre und Verbraucherschutz (zugegeben, klassische Grüne und FDP-Themen) z.B. kommen doch gar nicht vor. Beim Thema geistiges Eigentum geht es der Union auch nur um maximale industrielle Ausbeutbarkeit, die Probleme die z.B. Softwarepatente für den Mittelstand darstellen, oder wie sehr unsere Kultur (!) von der Privatkopie lebt ignoriert sie komplett. Beim Thema Internet ist sie sowieso noch 7 Jahre zurückgeblieben. Die Grünen nutzen es schon lange intensiv zur Kommunikation mit der "Basis" (auch ausserhalb der Partei).

Die FDP ist da schon sympathischer: diese Vertritt zu vielen der oben genannten Themen sehr ähnliche Punkte wie die Grünen. Wie da eine Koalition mit der Union gutgehen soll ist mir schleierhaft...

Am 13. September kommt übrigens Joschka Fischer nach München auf seiner Wahlkampftour. Als Aussenminister hat er einen exzellenten Job gemacht - Deutschland ist ein angesehenes Land, das wie kaum ein anderes für Frieden geradesteht! Das reduziert vermutlich unsere "Terror-Gefahr" mehr als alle Maßnahmen von Schili und Beckstein zusammen!

Neue Arbeitsplätze entstehen hier: in den regenerativen Energien und der ökologischen Landwirtschaft ist der deutsche Mittelstand führend!

Wouter Verhelst complained about the long user-agent string of Konqueror.

User agent strings are used by many web app writers to optimize their broken webpages for what they think your browser is. As I've already said in my previous blog post, most web app writers can't write proper code (nor proper HTML). So they have thousands of broken ways to deal with user agent strings.

Even Microsoft, who is really not a fan of Mozilla, claims to be Mozilla.

Including some information about your browser and OS can in fact be useful. E.g. if you visit the firefox homepage, it will provide a quick download link for your language and operating system. Agreed: Debian users don't need this download link. Still this is a good thing at first.

If you are concerned with privacy, you should probably use privoxy, or override the user agent string. But you aren't anonymous on the web anyway...

Back to user-agent strings. I recommend sticking to the Mozilla User-Agent string specs. They have collected some experience on what you can do to your user-agent string without breaking stuff.

But I do agree with you that the full Debian package revision string is maybe a bit extensive, while the "like Gecko" thing is stupid.

For those interested, here's the Galeon string: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050802 Galeon/1.3.21 (Debian package 1.3.21-6)

There is few of bogus information here: it is Mozilla/5.0, and the "X11; U; Linux i686; en-US" stuff is pretty much standard since Netscape 3 or so. The mozilla revision 1.7.10 is what I am in fact using, the engine is Gecko while the UI is Galeon (which e.g. means that I can use smart bookmarks).

And I think it's good to include Debian in the version string. This will increase our projects visbility. ;-)

I'm very pessimistic about software quality, as you might have gathered from my previous blog post. Especially when it comes to PHP software. I havn't had any issues, just a bad feeling with lots of software.

It's way too easy to write bad code in PHP. When you write bad code in C, it's likely to just not work at all. So noone will use your software.

When you write bad code in PHP and make it generate a fancy webpage, everybody will rush to use it. I've seen really bad stuff, e.g. a Fancy looking webmail interface which had the bad habit of storing all passwords ever used in a MySQL table - but never removing them. I discovered that only when I was to migrate it to a new server. When I was about to dump the mySQL tables I almost had a heart attack. (Obviously, the web server needs to be able to read the password to access this mysql file, so any slight issue in any PHP script on that box would have allowed the retrieval of dozens of passwords!) Fortunately, only a few different users of that server have been using this webmail, while most used real email programs or the other, older webmail interface.

Then there are the recurring issues with SQL injection vulnerabilities in PHP scripts, authentication bypasses (script.php?authenticated=1) and issues when executing imagemagick.

I was the maintainer of libming for quite some time, but the code of it was pretty much unmaintainable. Every now and then, extensions would be broken. Development was mostly stalled, and I bet the code was never audited. That was when I decided to orphan and have it removed 3 years ago or so. Since then I get like 1 mail every three months asking if I have updated packages somewhere. Granted, ming development has been picked up by others in the meantime. Still I have doubts that anything except the PHP module is working...

On the long run, we would need to audit lots of code in Debian. Right now, we are relying on upstream, commercial linux distributions and external companies to do that. But look closely: Novell released SuSE to become "openSuSE", RedHat has split of Fedora. They are not going to provide much additional security audits.

Maybe we should add a Debtags tag named "quality::audited-by-independent-party".

Hmm... and maybe I should stop toying with Python. It probably is as easy to write bad Python code as it is for bad PHP code... (except that PHP code usually is unreadable, because it's badly intermixed with HTML fragments)

And I also know lots of python code I have a rather bad opinion of... (e.g. mailman, or offlinesync, which has unhandled (!) exceptions when I have accessed my email on the server with mutt, reported a year ago)

[Update: Steve Kemp sent me an email, that he agrees with my view, and points me to the Debian Security Audit project, which could use a lot of help.]

Like Clint Adams, I disagree with the opinion of Philip Kern with respect to not allowing comments on blog postings.

I do not trust random PHP scripts on my webserver. I don't want my blog to have write permissions (especially since badly written PHP scripts is by far the most common intrusion vector). I don't want dozens of plugins from different authors (who may or may not be able write secure code) in my cgis. I want to upload my blog postings with ssh to my server. And no, comments won't work that way, so what?

Of course I do update blog postings with backlinks sent to me by email (you remember email? the thing everybody used before planets were invented, and when mailing lists weren't so high-traffic you need 4h a day to keep up with) when they are really useful. I don't claim my blog is "democratic" and about free speech, but only that it reflects my own opinion.

Yes, we are abusing the blog medium somehow by "talking" this way. And sooner or later (as more people are being added to planet.debian) it will suffer from the same problems as the mailing lists did: Too much volume. Then someone will probably come up with something new. Forums btw. are neither new nor a solution, they are even worse.

Btw, I can't remember that there is a requirement to have a debian account to be listed, you just need some DD to add your blog. Try setting up a Debian category, posting some useful stuff and then ask some DD to add you)

Ah, and yes, planet Debian is broken every now and then. For example it gives out incorrect permalinks for my blog posts (what is so difficult about isPermaLink="no"?). I have the impression planet was written without ever looking at the RSS specs. And without looking at the XML specs either, judging from earlier issues (e.g. breaking whenever anyone uses an ampersand in a posting title)

I've been a bit adventurous in learning new stuff. I've just done some small programs in Python, and found it a compact, powerful language.

I've also not done much with Gtk. Nor with glade, which is a pretty interface builder from the Gnome folks.

So why not combine these things to a new project?

So I started playing around with PyGTK.

Here's the current result:

This is around 180 lines of python code (250 lines total). Badly written mostly... ;-) but a lot of stuff is already working:

The data is loaded from an xml file (hooray, libxml python bindings), which is used to generate the tree on the left. When you select a node in the tree, the corresponding contents from the xml file (marked up with "bold", "italic" and "underline" elements) are displayed in the textview on the right. You can edit the contents, cut, copy and paste are working, as are the bold, italic and underline buttons above. They will also change when you move the cursor around.

Another 65 lines of code (that need to be rewritten, albeit they are working) go into refactoring the marked up text into XML, because the textview allows overlapping tags, XML doesn't. So if you leave this dead code away, it's just 120 lines of code!

When I'm finished, this will be something like tuxcards (except not in Qt), or CueCards (non-free, windows only, but has a freeware version my mother uses a lot)

So what's left to do:

• rewrite the to-xml formatter to use libxml2
• edit functions for the "pages" in the tree
• copy & paste for subtrees
• output (well, save the xml file)
• file dialogs

But it's amazing on how much you can do with just 120 lines of python code!

Update: I couldn't sleep. The code is now 259 lines total, 184 lines of code. It now updates the XML data and dumps it on exit. Saving is now trivial.