On Friday, a branch for GCC 4.2 finally got created, which means that we will hopefully see a release in a few months. The branch should have been created ages ago but the number of regressions just wouldn't go under 100 until recently. During that time, basically since I completed my tests with GCC 4.1, I've been busy testing snapshots of 4.2. I've mainly been testing on em64t (amd64), powerpc and ia64, but I also did some runs on alpha, mips, mipsel, s390 and sparc. Recently, I started testing it on ARM. Being an embedded architecture, ARM isn't terribly fast compared to some other architectures. However, Intel's IOP line is quite interesting and is used in a number of NAS devices. They typically include SATA and often have expandable memory. The device some ARM people are currently working with is the Thecus N2100. A port of debian-installer is underway but more on that later. Since my N2100 is not permanently connected to the Internet, Riku Voipio kindly gave me access to his to do some GCC tests. I started several weeks ago using gcc-snapshot 20060922-1 and it has been compiling happily since. I sort packages by their age, starting with old packages, so while there has been quite a bit of progress since I started, lately it has been going quite slowly. With about 2000 packages compiled in 3.5 weeks, I reckon that GCC 4.2 will be released before I've compiled the full archive. I've therefore been thinking of using distcc to speed the process up. The idea is to run the build process natively on the ARM box but use distcc to perform the actual compilation on another box, namely on a fast machine that has an ARM cross compiler. Unfortunately, I don't have access to such a setup right now. However, I strongly believe that this is a good alternative to test GCC on slower systems and in fact Bill Allombert is currently testing whether it could be used for the m68k port. Finally, Intel's new IOP 34x CPUs are also interesting for this kind of work given that they feature a cache and go up to 1.2 GHz with two cores. P.S. If someone is interested in fixing bugs, I have tagged package bugs related to GCC 4.2.

Cool, I'm heading to Spain for the QA meeting in Extremadura in December. Not quite so cool is that I'll be flying out two days after getting back from a holiday in Australia. I suspect I'm going to be totally shagged by the time I get back. $DEITY only knows what timezone my body will be in. Better get my QA pants back on again.

Gosh, I thought our own Republican gubernatorial candidate Kerry Healey had hit rock bottom with deceptive, racist, scaremongering television ads — e.g., this one (“What kind of person defends a brutal rapist?”) and this one (“Here’s a question: have you ever heard a woman compliment a rapist?”) (commentary). Hopefully it will backfire. But now, via the trademark blog, this in Tennessee. Update: and now this.

I just uploaded Audio::XMMSClient::XMLRPC 0.01 to the CPAN. It's basically an XMLRPC interface to the xmms2 daemon. It ties all Audio::XMMSClient functions except for the signal and broadcast stuff to an XMLRPC API which is quite similar to the Perl API of Audio::XMMSClient. It will be available from from CPAN in some hours and can be grabbed from the Pause incoming queue in the meantime.

For a while I’ve been looking for something to put under my laptop when using it on the couch or bed (it has vents on the bottom and runs pretty hot). I don’t have any large books, and the original idea to buy a cutting board for this purpose has failed spectacularly. All boards I could find in the nearby stores were either too small, or priced way higher than I was willing to pay. Determined to resolve this fundamental problem, I visited a local Home Depot today, and found an excellent solution: a 1′x1′ marble tile. It works beautifully, and is much more aesthetically pleasing than anything else I could come up with.

This seems to be a happy day for Debian. It certainly gave me the warm fuzzies when I read people's reactions to the Slashdot item, the steering committee, Madduck's and Joeyh's reflections on how things can work and some comments on IRC. It pleased me enough to get some new bloging infrastructure going in order to be able to chipe in.

There definitely is hope for Debian and change for the better is possible (and actually happens), as seen in the improved climate on IRC.

As with every change for the better, there is need of "positive crisis awareness" in order to reflect on possible changes, build momentum to overcome the status quo and implement changes in order to improve the situation step by step.

Crisis awareness is the awareness that there is a problem that needs solving. In contrast to negative crisis awareness, where one is paralyzed and unable to act on the danger and basically rolls over and dies, positive crisis awareness motivates to initiate change in order to avoid the crash and burn point and turn it into an opportunity of improvement, perhaps.

And exactly this is what I see emerge more and more: People who are no longer satisfied with the current situation and formulate thoughts on how to achieve improvement. A year ago we saw this with the hostile climate on mailing lists and IRC, now people realize that Debian has a lack of leadership, and that this is actually a problem.

We are still not the welcoming and friendly project we ought to be, but we are less hostile already and people know in what direction they want to move in the future (the more friendly one, thankfully). People have realized that hostility and pissing contests are evil and not a sign of l33tness, as some might have thought some time ago. But a culture change is happening, and if we stay on top of the issue we have a very good chance to not only omit the negative, hostile things we used to do but also come up with actively inviting and friendly things we can do.

The same is happening now (on a lesser scale so far) with the leadership area. The great thing is that if Debian takes up its issues and continuously improves there is hardly a limit to its potential. Not the first and strongest, but be the most persistent and patient will win in the end. And if we have incentive and urge to change we might just manage to keep up a continuous improvement process.

I think *that* incentive is the greatest thing we have to thank Ubuntu for.

Today I watched the "An Inconvenient Truth" film which was recommened to us at the foo camp preceeding Euro OSCon. Steve had already blogged about it shortly. While i enjoy the summer evenings at the end of september that I experienced this week, I think we are heading for very serious and even hard times. I would guess that we all will soon have to cut back on many areas of our compfortable lifes: less warm showers, more sweaters in winter, not keeping your computer on 24/7, no more car, no more flights. I have been thinking on where to move, given the delicate position that europe is in with its dependency on the gulf stream.

Incidently i also ran into this cool website today, which makes one consequence of global warming a little bit more realistic. If you are lucky your house in the middle of the city will be beach front property soon.

Yet another example of that positive crisis awareness is needed. Act today and we might just get a major global crisis. Don't act and we might get mass extinction.

This is the reverse lazyweb: I want to share what I learned, since many have similar considerations.

Some time ago I started to use the c't hotline to inform myself before buying computer stuff. Currently I am looking for a new notebook, since my wife poured tea into her old one and the keyboard stopped working.

So today I talked with one of the notebook experts at c't. My criteria were: "small", "long battery lifetime", "reliable", "good warranty", "excellent keyboard" and "works well with Linux".

He recommended that I should look at business notebooks as they came usually in higher quality and with better warranty. He pointed out a downward trend at Lenovo, which seems to produce lower quality machines nower days. Both warranty and technology seem to get reduced in quality. Asus machines break more often then average and their repair/warranty service is bad and improves only slowly. Dell machines break more frequently but their service is good. Linux compatibility seems good at Fujitsu-Siemens notebooks, which seems to have good service/warranty, too. Keyboards have improved a lot over the last few years and he encountered no bad keyboard at any decently priced notebook nower days.

I remember that joeyh and fabbe independently from each other picked a Fujitsu-Siemens. I will look into their offerings now.

I used to use the acpi-support package. It originally comes from Ubuntu. The problem is, it really does a *lot* of weird things when suspending and resuming. There are all sorts of workarounds for broken drivers and hardware. And in the end, all these hacks only made things worse: it actually worked better if I just did "echo mem >/sys/power/state" than if I used acpi-support.

I guess this is an Ubuntu thing, hack around everything and try to make it work. It gives fast results. In Debian, I think there's often a kind of opposite perspective: if the driver is broken, fix the driver, don't add ugly hacks. Both approaches of course have their pros and cons.

But now, I have switched to the hibernate package. I like it a lot better, because it isn't just a collection of shell scripts. You can actually configure which workarounds you want to enable, and how you want things to be done. I don't know why Ubuntu didn't simply use this package instead of inventing their own. Maybe hibernate wasn't around at the time.

In the long term, there really needs to be distribution-independent suspend/hibernate interface though. The current situation is less than ideal. Maybe pm-utils will fill this gap.

Update: Oh, I forgot. acpi-support is configurable, but not in a particularly intuitive way, at least not when compared with hibernate. Or maybe it's just my irrational taste.

Update: What I meant to say here actually, I think, is that hibernate is a lot easier to configure and much better documented. You can in a much easier way control exactly which workarounds/hacks you want to apply. And the options are described in the manpage for hibernate.conf. Editing /etc/default/acpi-support is a lot less flexible and a bit confusing.

Hmm, let's see if we can turn britney hint file comments into a blog entry... The apache 2.2 transition is coming along rather well. This is good news, giving us a shot at including a version of apache2 in etch with support for files > 2GB in size. Even though pervasive LFS support has been a goal in Debian since before sarge, when a big part of the interface your library exports to users deals with files it gets tricky to switch on LFS support because of ABI changes. After tagging for removal a number of apache2 modules with unrelated build problems, that leaves us with these issues blocking the transition:

• four binaries that need to be built/uploaded by autobuilders
• four packages with binaries waiting until next dinstall to be available
• one binary removal from unstable by a member of the ftp team
• libapache2-mod-auth-pam fixed for compatibility with the new Apache auth API
• libapache2-mod-auth-pgsql and libapache2-mod-auth-plain reviewed for compatibility with the same API
• fix a typo in mod-bt
• update medusa to pick up the new paths for libapr
• fix a gcc bug that keeps us from rebuilding pysvn

Nothing that a medium-sized hammer can't take care of... The release managers have also issued a statement about where we stand now for non-free firmware in etch. I'm sure this isn't anything close to the last word on the subject, but hopefully it will be enough to let us begin moving forward again instead of spinning our wheels on vote discussions.

It's been days since I've blogged -- I've got a lot to write about Halifax and I've been putting it off, but I wanted to jump on one more time to put in a plug for tomorrow's BarCampMontreal. I keep forgetting how great this city is for interesting technical people. Simon Law is going to be talking about how Web 2.0 is destroying Open Source software, and Austin Hall will talk about how you can get billions of dollars (American) with Zero Knowledge. The full Horaire shows the proposed schedule, but... hey, it's BarCamp, right? Anything could happen. As little as I like being up and about at 9AM on a Saturday morning, I'm excited about the event tomorrow. Should be a good time. tags: barcamp montreal barcampmontreal

I was kind of thinking of starting blogging again (after having blogged only two entries here). I also wanted my own blogging software; not this livejournal thing. I want to be able to edit my blog entries with emacs, the easy way, and keep it all in a git or darcs repo. I want to be able to choose my own editing language; whether it be markdown, textile, rd or whatever.

Of course I thought of Joey's nice ikiwiki. The only problem is that I don't like its commenting system. I want people to be able to comment easily. I like comments. It is probably possible to extend ikiwiki to make it easier to do comments, but it's not something I want to do right now.

I took a look at this weird thing called hobix. I liked the design and the fact that it is written in ruby, but the author seems to be on LSD. And all this blahhrhrg whsyyythe thing, though marginally amusing, is not the kind of solid documentation I'm looking for.

Yes, I know, there are tons of blogging engines out there. I just don't _like_ them. I hate PHP, and I hate SQL databases (with a possible exception for SQLite). I just want something simple, dammit.

But I'm tired now, and my neck hurts. I will look more another day.

Update: I forgot to mention that I've tried pyblosxom, and it really looks most similar to what I need. I just thought it was a bit complicated to set up with the features I wanted. But maybe it was just me being lazy.

Although I like the idea of a tiling and completely keyboard focused window manager, I never fell in love with Ion because the default keybindings weren’t really intuïtive (to me). A few months ago I noticed, that ratpoison is also a tiling and completely keyboard focused window manager, only with much more intuitive usage: If you know screen and it’s keybindings, you also know ratpoison and it’s keybindings: Just exchange Ctrl-A with Ctrl-T. This sounds perfect for usage on my low performance laptops, where I have small screens and usually also no virtual desktops in use. There’s only one thing which annoys me in ratpoison: If I use a mostly mouse driven application like e.g. a webbrowser with ratpoison, I have no problems to click on links, even if the webbrowser is not in the so called “current frame”. But if e.g. click into an input field, I usually notice much too late that while the mouse works fine in the browser, keyboard focus is still in some other window. Currently they all use flwm, the Fast and Lite Window Manager. So what I would need is a tiling and keyboard focused window manager but with “focus follows mouse” politics. And since the laptops on which I intend to use such a window manager, all have a touchpad or thumbstick, the mouse there counts as keyboard focused, too somehow, doesn’t it? :-) I wonder, if an ion3 could be configured to use the same keybindings as ratpoison. That would probably fulfil this desire. On the other hand, there are browsers which are fine without mouse. lynx or links2 for example, so the focus problem I have with ratpoison wouldn’t occur. But what if I need or want a keyboard driven and full blown webbrowser? Ok, Firefox as well as Opera are not that bad in keyboard only use, but they still are focused on the mouse using user. But Gecko wouldn’t be Gecko, if there wasn’t some Gecko based browser with this features: On the ratpoison website I found a link to a very interesting Firefox plugin which makes Firefox a complete new browser, a keyboard driven webbrowser named Conkeror. It has no toolbars at all, no (visible) tabs, no menus, no nothing — it shows only the website in fullscreen, a status line and a multipurpose command line — exactly like the mini-buffer of GNU Emacs. But not only the layout, even the keybindings are very emacsish: C-x C-f opens an URL in a new buffer -eh- tab, C-x 5 C-f opens an URL in a new frame (window), C-x C-v opens a new URL in the current tab (buffer) with the current URL as editable default value, C-x b switches to another tab, C-x k kills -eh- closes a tab, C-x C-b lists all open tabs, l goes back (remember the Emacs info reader, eh?), C-g quits accidently requested dialogs or stops loading a web page, Ctrl-s and Ctrl-r give you forward and backward i-search, C-n, C-p, C-f and C-b scroll, etc. Even M-x works, e.g. will M-x revert-buffer reload the web page. (Unfortunately Esc-x doesn’t work. Yet.) And for vi freaks, there is even M-x use-vi-keys. There’s even one lynxish keybinding: \ lets you view the source. And although it’s one of the strangest webbrowsers I saw yet, I somehow like it and also would like to see it in Debian as package, since it is the perfect companion for ion or ratpoison. Looking through apt’s package cache as well as the wnpp bugs, I haven’t found any hint on somebody already packaging it, so I’ll have a look on it and on how to to package a Firefox extension for Debian. BTW: While looking through the wnpp bugs, I found bug #335459, which is the ITP flock, an also Gecko based browser with a lot of cool features for blogger who like social network tools. Another nice thing I found today in Debian was the xfonts-artwiz package whose small fonts are very suitable for small resolution screens, especially if a tiling window manager is used with a e.g. 800×600 resolution. Unfortunately they aren’t available in a charset with German umlauts. Apropos tiling window managers: Anyone tried pconsole with an automatically tiling and resizing window manager? I wonder if it’s usable. At least on MacOS X with its cascading window positioning algorithm, pconsole is a pain. — But even without cascading windows, MacOS X is a pain for keyboard users. Just think of its default behaviour when using the tab key inside a form mask: It will skip all buttons, all checkboxes, all radio buttons and all select boxes. Argh!

Toyota in the US has released a hybrid Camry which seems to be the larger Prius that many people have wanted. Since it's release the Prius has been greatly desired by people who like technology and the environment. The only down-side to the Prius is that it is a small car and doesn't have as much room for baggage or passengers as you might desire.

The new Camry Hybrid has the Continuously Variable Transmission (CVT) that gives the Prius it's smooth ride, the keyless start (optional), DVD navigation via a large screen, six airbags, and tire pressure monitoring.

The above review states that the test car shuddered when the engine started and made odd mechanical noises. It's not indicated in the article but I suspect that the car may have been an early production test model rather than the type of car you will get if you actually buy one. The Prius I drove did not shudder or make any odd noises (in fact hardly any noise that I could hear). Maybe if I drove a Prius in a country area I would hear some noises, but in the city (which the Prius is designed for) the Prius didn't make any
engine sounds I could here. I expect that the hybrid Camry will perform as well as the Prius in this regard.

The review describes the hybrid Camry as giving 35mpg with a driving pattern that was not the most fuel efficient, that is roughly 6.7 liters per 100Km which is more than the quoted rate for some small cars. The four-door Smart cars are advertised as using 5.8L/100Km and Diesel allows even better efficiency. But if you want the space of a Camry then 7.1L/100Km is fairly good, especially considering that you could reduce that by driving more slowly. Also we have to consider that the primary aim of the Prius design (and presumably the design of the hybrid Camry) is to protect the environment by reducing the poisonous emissions, this may reduce the fuel efficiency slightly (the aim of reducing emissions is not always compatable with the aim of reducing fuel use).

There are apparently 3.785 liters to a US gallon. A US mile has 1760 yards and a yard is apparently 0.9144m, so a US mile would be 1760*0.9144 = 1609 meters or 1.609Km.

3.785*100/1.609/X

To convert MPG (the US measurement of fuel use when based on a US mile and a US gallon) to L/100KM (the Australian way of measuring fuel use) you use the above formula (where X is the MPG rating). I included this information here because I couldn't find it anywhere else.

I was going to post this before but was side-tracked by the flash issue.

Also I've included a beard picture for day 12 (yesterday).

It seems that yet again Microsoft developers are incapable of keeping time correctly. Apparently there's a bloody stupid bug in Exchange Server which means it will get the date of the end of BST (British Summer Time) wrong this year. October is only supposed to have 4 Sundays, not 5. At least, that's the assumption their high-quality software makes. This is far from the first time that I've heard of silly bugs in Windows and Microsoft apps when it comes to handling time and date information. Still, it could be worse. At least they wouldn't try to bury this kind of insanity inside a programming language or inside an international "standard". Oh, wait...

Devil's Pie (someones favourite window manipulation tool) 0.18 is out. Features galore in this release!

• Add window_property matcher (Nigel Tao)
• Add set_viewport action (James Willcox)
• Add opacity action (Pavel Palat)
• Add sticky action (Steve Leung)
• Add unmaximise action (Alex Menk)
• Add unminimise action (Richard Neill)
• Fix negative offsets in geometry action (Vincent Ho, Midred)

Downloads are in the usual place.

Devil's Pie (someones favourite window manipulation tool) 0.18 is out. Features galore in this release!

• Add window_property matcher (Nigel Tao)
• Add set_viewport action (James Willcox)
• Add opacity action (Pavel Palat)
• Add sticky action (Steve Leung)
• Add unmaximise action (Alex Menk)
• Add unminimise action (Richard Neill)
• Fix negative offsets in geometry action (Vincent Ho, Midred)

Downloads are in the usual place.