Category: Uncategorized

The best thing about linux.conf.au so far was yesterday’s weather. It was mild with a slight breeze, and the venue has made the mode of it by using UNSW’s glass-walled Pavilion near the Clancy theatre for hack and chat space. (As with OSDC though, people can’t tell the difference between someone noodling on their laptop and someone doing work, and interrupt both sorts.)

Jeff gave the conference opening talk and forgot to introduce himself. It was a fun talk. About a week ago he called me to ask what he should say about the lca Woman Triumph. Over 10% of registrations were women, which doesn’t sound much until I remember that at linux.conf.au 2004 there were eight women — not eight percent, eight. I actually advised him not to dwell on behaviour too much; lest the men be insulted and the women think that they’d accidently signed up for Barbarian Finishing School (how to talk to girls, level 1). He did stick a don’t be creepy on the end of that section, with the result being that a couple of guys came up to me and joked about how they could be creepy but weren’t allowed to.

Guys: creepy is creepy, but meta-creepy is both stupid and irritating. Don’t do either of them, and don’t start on the meta-meta, thank you.

The first two days of the programme, yesterday and today, are mini-conferences (think workshops), organised by communities like Debian or GNOME, or by interest groups, like people interested in education. I am running a LinuxChix one today. The conference has decided to embrace them this year: rather than describing them as a sort of a pre-conference bonus, the mini-conference schedules went into the main schedule, and Chris Blizzard’s keynote is scheduled this morning to lure people to the mini-conferences afterwards.

Yesterday this seemed to be a bit of a mixed blessing, because the mini-conference organisation and programmes largely aren’t up to the standard of the main conference, which has a programme I just can’t stop drooling over. Two things! In every session! That I want to attend!

Andrew said yesterday that he should have stretched a bit and gone to the embedded miniconf rather than GNOME, and I sort of agree for myself too. I tried Education in the morning, but Pia had had to move her talk at the last minute, and in fact backed out of her slot later in the day too, leaving their programme in disarray for the casual attendee, because every session they’d announce a random talk.

I went to the GNOME talk on Avahi, which just didn’t grab me somehow, and then Jeff’s talk, entitled Connecting the dots, in which he mostly demonstrated the capabilities of his Wii. Now, actually this was kind of interesting for me, because Andrew and I didn’t set up our Wii until last night, so we’d never seen it in action. And James, who is well out of console-land, was fascinated. But by the conference’s core criteria — talk about Free Software, or Free Culture — it was a complete disaster, since the Wii while cute, is extremely proprietry.

Since I had my own mini-conference today to prepare for, I didn’t attend any more sessions after session 3 yesterday. Instead I sat in the nice open air pavilions and worked on slides, and then Andrew and I took off for a yoga class in the evening and went out to dinner afterwards.

As an update to my note about advogato.org’s mooted closure, the new maintainer Steven Rainwater emailed me to let me know about the all new, inclusive Advogato: they’ve added an aggregator. If you have an Advogato account, you can now return to the recentlog by going to your account settings, ticking the Syndicate your blog from another site? box and then putting an RSS or Atom feed in.

Just spreading the word, since neither Andrew nor I received an email invoice for our registration: credit card payments for linux.conf.au are now being accepted. (We did get the announcement, but previous conference experience—ACL, HCSNet—this year has unfortunately taught me that registration information is not sent out using titles like Countdown to linux.conf.au 2007: 48 DAYS TO GO, those kind of titles now indicate to me we’ve updated the website! and now we have a directory of attendees!, ie, not action items. So, I didn’t actually read it. Oops.)

People who got the earlybird price (which closed Nov 15) must pay by December 8. You can also still register now and get the regular price, although as the announcement (also) pointed out, if you want to stay on campus in the pre-arranged accommodation, or you want to go to the dinner, register soon. (It’s unpredictable how full they really are, until they start re-opening spots that people haven’t paid for. But 450 attendees who haven’t put down money yet is still a goodly number when I believe the aim is 800.)

Dear universe, Andrew and I are in Melbourne for OSDC from Tuesday December 5 to Sunday December 10 (in the morning, anyway). If any of you are in Melbourne that week and would like to meet us for breakfast/lunch/dinner/drinkies, get in touch.

1. Registration is open, starting from the low low price of $99 for students and proceeding to $300 for self-funded attendees and $690 for professional attendees
2. the LinuxChix miniconference programme is available in draft form (waiting for some speakers to confirm that they can make it); and
3. if you’d like partial reimbursement of registration costs, you can volunteer to help out.

A relatively idle thought after doing Ubuntu support on LinuxChix lists for a while: are the code names really such a good idea? People have an enormous amount of trouble correctly identifying their Ubuntu version. I’ve seen the following problems:

1. people not realising that the zeros are significant in the version number and asking for support with Ubuntu 6.1 (they probably mean 6.10/Edgy Eft) or 6.6 (they probably mean 6.06 LTS/Dapper Drake);
2. at least half the time people quoting the Ubuntu version number and codename together quote a mismatched name and number (Ubuntu Breezy 6.06, Ubuntu Dapper 6.10 and that’s not even getting into Ubuntu Breezy 6.1 or Ubontoo/Urbanto/Obonto Dragon and so on), which means that you have no idea which version they actually mean; and
3. the code names are memorable, but seemingly not memorable enough, there’s a lot of people out there talking about the Edgy Elf, which sounds like a bad drug pusher.

Ubuntu is far from the only software using well publicised release code names. I remember the good old days pre-Windows 95 (the good old days are always more than a decade ago), when you couldn’t talk computers without talking about ‘Chicago’. Debian’s release code names are also very commonly used; potato, woody, sarge, how well I remember thee, and I have no idea what thy version numbers were. In fact, the problem might perhaps be that the release code names and the version numbers are essentially equally well known when it comes to Ubuntu, so people feel the need to state both and aren’t clear on the mapping between them.

I suspect also the regular releases are hard on people: people know that there’s lots of Ubuntus and they have to identify their one, but there’s changes often enough that casual onlookers and users are more confused by the release names than they are aided by them. The release numbers map to the release date (4.10 was released in October—month 10—of 2004, 5.04 in April 2005 and so on) but most people, I believe, treat version numbers as Marketing Magic the like of which mortals do not ken and question no further. The six month release cycle means that the current system always has several easily confused releases too (you can confuse either the first number, mixing up 5.04 with 5.10, say, or the second one, mixing up 5.10 with 6.10).

I don’t have any particular suggestion about an alternative, and suspect that the developer community is wedded to their names even if the users are a bit puzzled. I suppose simpler would be better: Ubuntu 1, Ubuntu 2… but then the numbers get high quickly.

Really final, because last time I hadn’t upgraded a server, and yesterday I did. It made me very sad indeed.

Original complaint	Bug number	Fixed?	Remaining sadness level
apt crashes when upgrading courier-authlib	64615	No (there are a couple of potluck workarounds in the report)	High, because it took me about an hour and a half to hunt down the bug report and get apt and dpkg to dig themselves out of the mess they were in.
Transparent proxying in Squid is broken	68818	No (there’s a fix and workaround in the upstream report)	High, because it took Andrew about an hour to hunt down the bug report and jigger with the upstream workaround. Yes I know transparent proxying is evil, and someday someone will figure out a clean way to autodistribute proxy settings whenever I connect my laptop to a new network, but until then I use it.
Network Manager can’t always tell the difference between wired and wireless cards	59981	Yes	Moderate, fixing this one has just made the intermittant appearances of 40125 more obvious, but somehow I find 40125 less irritiating.
Nevow is completely useless, won’t even import properly in Python	61423	No	Moderate. It turns out (and by it turns out I mean I figured out) you can install the Nevow 0.9 package from Debian unstable and it will work just great. But if Nevow isn’t supported even to the point of shipping an importable Python package, why is it still in Ubuntu’s main? This enhances my sadness level.
If you type words into the address bar, the epiphany browser no longer treats them as search terms for Google, it instead treats them as a bad URL.	56610	No (patch is available)	Low, since I was able to build the fixed package as suggested in the report.
When I attach my Canon IXUS 65 to my computer via USB, Import Photos reports Could not claim the IO device.	64146	Yes, or at least they say so and it works for me, but people are still adding to the bug.	Low, and only because I keep getting the bug mail about it.
X can’t always… work [actually, probably a bug in vbetool, causing rendering glitches on resume from suspend]	60882	Officially no, but I see it occur way less often now, maybe not at all since Edgy released.	Low, since I see it so seldom. It’s really annoying when it does pop up though.
Aptitude… is now incredibly slow to resolve dependencies	51893	Yes	None
GNOME reports that CPU scaling is not available on my machine	Wasn’t one, I didn’t have powernowd installed	N/A	None

There are various ways speakers can embarrass themselves at conferences by projecting personal information or embarrassing stuff onto a screen in front of many people. Background images featuring non-professional levels of nudity are the most well-known problem, but others include accidentally popping up a personal chat session, interesting titles being shown in your browser history, the contents of your clipboard being more informative than you’d like, that kind of thing. You don’t only need to protect yourself against overly sexual material (well, unless you’re presenting to the BDSM community or something), there’s also password leakage, and just the vague embarrassment of showing professional peers a log of you and your husband discussing whether or not to have pork for dinner.

It occurs to me that with a bit of discipline about files, the easiest way to avoid this on laptops where you can set up a second login is to do just that: have a second account which has suitably bland background, in which you only access presentation related websites, in which you don’t IM and so on. Set the password to something totally unrelated to your usual passwords, and change it immediately after each presentation, just in case the worst happens and you type it in such a way that it appears on the screen.

And then do all presentations logged in as your special cleaned up for the general public laptop user. If you don’t even want cross-presentation pollination you could re-create the account for each presentation or have a script that does so for you, just to be sure that there’s no browser history and such.

Original complaint	Bug number	Fixed?	Remaining sadness level
If you type words into the address bar, the epiphany browser no longer treats them as search terms for Google, it instead treats them as a bad URL.	56610	No	Moderate, although I only realised that this was a bug, rather than the spinning wheel of dubiously user friendly UI changes, yesterday.
Aptitude… is now incredibly slow to resolve dependencies	51893	Yes	None
GNOME reports that CPU scaling is not available on my machine	Wasn’t one, I didn’t have powernowd installed	N/A	None
Network Manager can’t always tell the difference between wired and wireless cards	59981	Yes	Moderate, fixing this one has just made the intermittant appearances of 40125 more obvious, but somehow I find 40125 less irritiating.
X can’t always… work [actually, probably a bug in vbetool, causing rendering glitches on resume from suspend]	60882	Officially no, but I see it occur way less often now, maybe not at all since Edgy released.	Low, since I see it so seldom. It’s really annoying when it does pop up though.
Nevow is completely useless, won’t even import properly in Python	61423	No	Moderate. It turns out (and by it turns out I mean I figured out) you can install the Nevow 0.9 package from Debian unstable and it will work just great. But if Nevow isn’t supported even to the point of shipping an importable Python package, why is it still in Ubuntu’s main? This enhances my sadness level.
When I attach my Canon IXUS 65 to my computer via USB, Import Photos reports Could not claim the IO device.	64146	Yes, or at least they say so and it works for me, but people are still adding to the bug.	Low, and only because I keep getting the bug mail about it.

Anyway, on the balance it worked out that Edgy is reasonably stable. Still no Breezy, but as stable as Dapper on my desktop. And I am asking for trouble playing with Network Manager. How I wish to smite things when stuff that used to work just fine stops for a while though. And I’m not using it on my servers: one of them is currently running a four month experiment and I don’t want to reboot it (the experiment is interruptable, but it takes about a day to recover from interruptions); and the other one runs a couple of Nevow sites and I’d prefer to use a supported package. Or at least supported.

Back in 2004 I finally noticed a really common construction that so far seems confined to American English: it’s not that big of a deal and similar (How big of a perv is Foley?, How big of a problem with security?, How large of a volcanic eruption would it take to cover the entire world with volcanic material?) In Commonwealth, or at least Australian, English this would be it’s not that big a deal (likewise how big a perv is Foley?, How big a problem with security?, How large a volcanic eruption would it take to cover the entire world with volcanic material?, although I suspect the last one would often be rephrased).

I asked about it at the time (and of course one never blogs alone) and completely failed to notice that the answer came from on high, that is, that Mark Liberman from the Language Log bothered to answer. Wow.