Search Results: "ag"

31 March 2007

Joey Hess: random beauties

30 March 2007

Simon Richter: uClibc

The mighty uClibc has been repackaged and should be on its way to the NEW queue soon. The most prominent change is that it can now be cross-compiled for a number of architectures using dpkg-buildpackage's -a switch (you might want to install dpkg-cross). There are a few regressions that will have to be sorted out by trial and error, but these only affect architectures no one cares about anyway, like i386. Known bugs: There is probably a lot more that will only crop up when actual other packages are built against it (so far, my shiny new busybox works, but the CFLAGS I had to pass to make that work are illegal in some states). Next up: reintegrating the magic specs file that implements (cpu)-linux-uclibc.

Patrick Winnertz: cipux as admintool for DebianEdu - a second test

Mh… I tested today again cipux, a admintool which one of the two tools which will replace our(Debian-edu) old admintool wlus . Since I dislike cipux a bit I regulary test it (the last time 1 week ago) and most of the errors I announced last week are fixed. But here the new report: The CipUX team fixed many issues I mentioned the last time… but there are still very hugh errors for productive software :S During the installation an the run of cipux_setup -s I recognised an error in the script: it uses wrong options for an other subinstallation script of cipux. The cookie handling is mostly fixed but in some scripts I still got this nice picture.. Apparently there are still hugh errors in the cookie handling. Furthermore it was not possible to edit the user data (e.g. full name, …. ) in the useradmin, it allows only to add users, resetting the passwords and deleting them. I think this is not enough for an admintool as we need it in debian-edu as a replacement for webmin/wlus. While searching for editing the users data I clicked on “Userselfadmin” and … got a error 404.. this script does not exist. So here is also no possibility to change the user details. I noticed something strange… there are many groups for users, e.g: assistant, teacher, tutor, … but I don’t find an group for students, quite strange. I asked upstream but until one week I got no answer. :S Because I was a bit frustrated I want to have a look on one of the other features of cipux.. the internetadmin. It should be possible to forbid a user or a complete group the internet access…this is the theorie. When I clicked on the internetadmin-button I got the cookie error mentioned above. Mh… so it is not possible to forbid your pupils to go into the internet. :S Another important feature we need for DebianEdu. Okay… the other things works. ;-) As a conclusion: Upstream works hard on fixing the errors I found last time, but there are still hugh errors which can be found by testing the software a bit. Atm I would suggest them to work on it and don’t upload to Debian-Edu since I think many features they have in theory doesn’t work properly and are not well tested.

Patrick Winnertz: cipux as admintool for DebianEdu - a second test

Mh… I tested today again cipux, a admintool which one of the two tools which will replace our(Debian-edu) old admintool wlus . I started last week to test it and most of the errors I announced last week are fixed. But here the new report: The CipUX team fixed many issues I mentioned the last time… but there are still very hugh errors for productive software :S During the installation an the run of cipux_setup -s I recognised an error in the script: it uses wrong options for an other subinstallation script of cipux. The cookie handling is mostly fixed but in some scripts I still got this nice picture.. Apparently there are still hugh errors in the cookie handling. Furthermore it was not possible to edit the user data (e.g. full name, …. ) in the useradmin, it allows only to add users, resetting the passwords and deleting them. I think this is not enough for an admintool as we need it in debian-edu as a replacement for webmin/wlus. While searching for editing the users data I clicked on “Userselfadmin” and … got a error 404.. this script does not exist. So here is also no possibility to change the user details. I noticed something strange… there are many groups for users, e.g: assistant, teacher, tutor, … but I don’t find an group for students, quite strange. I asked upstream but until one week I got no answer. :S Because I was a bit frustrated I want to have a look on one of the other features of cipux.. the internetadmin. It should be possible to forbid a user or a complete group the internet access…this is the theorie. When I clicked on the internetadmin-button I got the cookie error mentioned above. Mh… so it is not possible to forbid your pupils to go into the internet. :S Another important feature we need for DebianEdu. Okay… the other things works. ;-) As a conclusion: Upstream works hard on fixing the errors I found last time, but there are still hugh errors which can be found by testing the software a bit. Atm I would suggest them to work on it and don’t upload to Debian-Edu since I think many features they have in theory doesn’t work properly and are not well tested.

Rob Bradford: Dates translations

We will be making a release of Dates soon and so are on the look out for more translations. To submit a translation please file a bug on the OpenedHand Bugzilla against the Dates product. The latest code can be found in trunk.

Andrew Pollock: [tech] DOWNTIME: lists.linux.org.au and linux.org.au email

What
migration of lists.linux.org.au and linux.org.au email to new server
When
Saturday, March 31 09:00:00 GMT-7 - Sunday, April 1 17:00:00 GMT-7 (Sunday, April 1 02:00:00 GMT+10 - Monday, April 2 12:00:00 GMT+10)
Contact
admin@linux.org.au (oh gawd, not more spam)Caesar on #linux-aus on Freenode
Details
I'll be performing a move of all of the Mailman lists for lists.linux.org.au from the hastily slapped together copy of digital running inside of a Xen instance hosted by IVT (thanks, Jon!) to a brand-new installation in a Xen instance called morton.linux.org.au on the hardware donated to Linux Australia by HP sometime after linux.conf.au 2006. I'll also be moving linux.org.au mail and local mailboxes from digital morton. All shell users of digital have already had their home directories migrated from digital to morton and been individually notified. I don't expect the outage to be anywhere near as long as specified, but I'm reserving that window of time for any problems, rollbacks, troubleshooting, etc. I'll monitor #linux-aus after I've moved the MXes and enabled mail delivery on morton, for any issues that arise.

Evan Prodromou: 8 Germinal CCXV

Had a great time at DemoCampMontreal2 tonight. There were a ton of people there, the demos were great, and the bar at the SAT was open for business, so it was possible to drink beer and watch demos. There were all very solid projects with lots of interesting stuff going on. My favorite aha! moment came from Hugh McGuire showing off Collectik, his new social-podcast-aggregation platform, when he demo'd how he uses the site, and I realized that that's how I want to organize my podcast listening, too. I also liked Kakiloc, a location-oriented social networking site. They let you tell your friends what you're doing and where you are, and the friends can see it on the Web, by SMS, or through IM. Sound familiar? Yeah. Kakiloc's advantage is that they have more geo-oriented interface, and some extremely powerful rules that trigger activity when you (say) get close to someone in your circle of friends. However, I think in a world of Worse Is Better Kakiloc is going to have a hard time making a sale. Twitter already has the mindshare in this market, and the network effect makes it more valuable for people to stay with the popular platform. I also really dug the films from OpenSourceCinema.org, although they skirted the "demo" concept pretty scarcely. One thing that was neat was that the OSC Web site was built by a real-life barn-raising: Sylvain Carle put out the call for some help, and people convened on his house to build the Web site. Cool idea. All in all I had a really good time. DemoCampMontreal2 worked out well. tags:

Bait and switch So, on the way out the door to DemoCamp, I grabbed a handful of loonies out of my bedside change basket for the parking meters downtown. When I got to SAT, though, I realized that the "loonies" I'd brought were in fact Sacagawea dollars. I'd gotten them as change from the ticket machines on the Las Vegas monorail. Doh! Fortunately, downtown Montreal parking meters take credit cards, so I was OK. tags:

Feisty Upgraded to FeistyFawn on Wednesday. Worked fine, nice smooth upgrade (who expects that?). Nothing revolutionary, but the software works well. tags:

Adam Rosi-Kessel: A Long Swim

Via UH, Google Maps Driving Directions from Chicago to London. Myself, I can barely tread water.

MJ Ray: My DPL Vote, but While There's Moonlight and Music and Love and Romance...

A lot of people seem to post their votes onto planet these days. More so than 2 or 3 years ago. I'm still not going to post the full details, but here's my usual summary, based on the debate and the platforms.
Above None Of The Above
  • Sam Hocevar - I share his view of the problems, although I don't agree with all the suggested actions. A bit quirky, but seems nice enough.
  • Aigars Mahinovs - I think he could be an accountable and fair DPL. Has ideas - not all good ideas and I don't like the aim of releasing less often - but I can't see him driving on if he's getting flamed to a well-done crisp. A leader more than a ruler.
  • Gustavo Franco - nice guy, but maybe I just think that because I've had more direct contact with him than other candidates recently. I agree with most of his views of the problems, except Ubuntu and the tattoo. Maybe too inexperienced for this time.
  • Raphael Hertzog - I think the team model is a good idea for the size of the project now and it learns from past attempts.
Borderline
  • Wouter Verhelst - I'd be voting mainly based on what I've seen of him. I don't want to vote for someone standing with a low-detail platform, as that's just scary. Even when people break promises, it's hard enough to remove them. The debate did give a bit more detail. I fear the adoption of ESR-style "authority follows responsibility" - that's just impractical for parts of the debian infrastructure.
  • Simon Richter - Generally OK, but I'm worried by the cutting it fine with the platform, not posting any rebuttals and being late for the debate because one battery went flat.
Below NOTA
  • Steve McIntyre - This won't surprise anyone: it's because of the manner in which the Debian UK Society retail business was started; but also the trial misadvertised as mediation of the d-i conflict. I shudder to think what social conformance tests he wants in NM.
  • Anthony Towns - last year I didn't vote for this politician (while correct in one way, most DDs have never even seen me in a mediation situation and he didn't have the courage to cc me on that email) because I thought that email suggested he'd be a terrible DPL. This year, he's demonstrated how terrible and the platform and debate suggests he hasn't even noticed: he thinks he's been a successful leader, despite the recall vote. One or two interesting ideas, but not enough to overcome that. His rebuttal seems to utterly miss Sam Hocevar's point about how being not-evil (Corporate Social Responsibility, as seen in the private sector) is different from actively being good (Concern for Community, as seen in the third sector): is that being dense or devious?
So, how wrong am I this year? Is there more trouble ahead?

29 March 2007

Alexander Schmehl: How do you notice...

...that the distribution running on your server is really stable? After several years you suddenly notice that reportbug is not installed. And you just wanted to report a wishlist bug...

Romain Francoise: Lessons learned: #612

If you admin a heavily loaded SpamAssassin server...
  1. Use SQL for Bayes and AWL data. This allows your spamd processes to store their tokens simultaneously, instead of fighting over the lock for the Berkeley DB files. I use MySQL, postgres works too.
  2. If you use network tests, install a local caching nameserver. Even if your nameserver is on the LAN right next to the mail server, it'll be worth it.
  3. If you use OCR, make sure that the caching mechanism works. This is especially important in corporate environments where people use images as signatures. For extra points, store the image hashes in SQL too.
Also, the old prefork (--round-robin) algorithm works much better in my case than the new Apache-like scaling algorithm. YMMV.

Arnaud Vandyck: jboss-hibernate in Debian

Today, I first fix debian/copyright file in libjazzy-java as it was rejected by ftp-masters. Then, I started to work on packaging jboss-hibernate in Debian. Marcus Better already filed some ITP's and I'm using them, change the owner when I'm ready to work on a package and upload it. #385569: ITP: jboss-hibernate -- object/relational persistence service for Java The good thing with the way Marcus worked is he filed ITP's and then send 'block' instructions to BTS, so it's easy to see what should be packaged first:

Benjamin Mako Hill: Dollar Books

One of my favorite weekend activities is spending a Saturday afternoon going through the dollar book carts outside many New York and Boston area used book stores. It's not only because the books are cheap -- although I like that. The dollar book section is the great equalizer of bookstores. A neglected Dickens or Shakespeare can rub out against a discarded Mary-Cate and Ashley teeny-bopper quasi-romance novel. In the best cases, random shelving creates perfect (if unintuitive and ironic) pairs like Run Run Run about Abbie Hoffman and Fun Fun Fun about other youngsters -- a pair I found adjacent in a dollar book section last week.
/copyrighteous/images/run_fun_run.jpg
But you also get to meet books you will never meet in the bookstore sections you normally frequent. The dollar book section at the Strand has introduced me to a whole class of books with intriguing non-fiction sounding titles that I had pulled excitedly off the shelf only to find out that they were, in fact, novels. There are also the books with titles so good you suspect the book will be downhill from that point. The Complete Idiot's Guide To Being a Model (not, as one might expect, Becoming a model) and this guide to the British pierage are great examples.
/copyrighteous/images/british_piers.jpg

MJ Ray: Google Bugs

A relatively recent addition to the Google range is Google Docs. I'll be testing it shortly, but I see it has already suffered google's usual security/privacy bugs Apart from encouraging the larger Atom feed format, dropping RDF support and unreliable alternatives to its eyetests, Blogger is one of Google's best services. The basic service seems to work in a half-decent manner. Of course, it still has bugs and that let someone post a hoax Google news release Google's search results are still unstable: here's a consultant spotting lots of turbulance in the results. If Google's actions were unfair and harmed your performance, you can't touch them, as kinderstart discovered (but that ruling might be an interesting reference for other ratings sites, so there's a silver lining). The site complains constantly that I don't let it rampage through my browser and I haven't downloaded the latest from Adobe. It's a bit less annoying now that ytplay is available, but it shows that Google doesn't care much for free software support. An interesting twist has been Viacom suing Google for a billion dollars and calling them all sorts of names. (Beware bias in the BBC reporting on this, as BBC just signed up for Google-YouTube [BBC] - is there a monopoly that the BBC won't befriend?) Adwords and adsense looks like some of Google's top mass-market money-makers. I have mixed reports about them, including some of my customers that are happy to use them, and others that get results like this commenter at Problogger which suggests that adsense referrals maybe don't always report accurately. There is also a privacy/trust problem with having one company seeing so much of the world's searches. This was the basis for a recent Ask.com ad campaign (nice idea, shame about the microsite's poor accessibility and various other things - it gave one Google-worker a good laugh and got better commentary from Susan Hallam Communications). Privacy bodies backed Google's step towards allowing user privacy [BBC] but there's more that should be done. That's enough for now. I've got some more positive news which I'll post along with the Docs review.

Pablo S. Torralba: ADN - Everything is connected

That's the company's slogan, whose offices we are working at. And I must admit it seems quite true.

Today, we have visited one of the data centers we are evaluating for our company hosting. It was a Tier 1 company, that is networking, connections... nothing special there. The amazing thing is one of the people we meet there was visiting Australia a year ago, and he is also a photograph passionated. I feel him a bit like a twin soul of mine as we are planning to visit Australia on our marriage trip where I will take a lot of pictures, of course. We were talking and he gave me some interesting URLs about his blog, travel agency and the like. So again everything seems connected, as Google knows so frightening well.

Anyway, we went there by motorbike which was a plus not connected to anything else, but too much funny in any case.

Matthew Garrett

I'm not really sure which I find more disappointing - this, this or this.

The Ubuntu Code of Conduct is something that's designed to ensure two things. Firstly, and perhaps most obviously, that discussions don't end up bogged down in personal abuse with no useful conclusions being drawn. Secondly, and perhaps more subtly, to ensure that potential contributors aren't put off by an atmosphere of perceived hostility.

The second point has been most commonly raised with respect to women in computing (perhaps most notably by the FLOSSPOLS findings), but it's not limited to that. One of the reasons I'm not involved with Debian nowadays is the atmosphere on the mailing lists[1]. If you're about to make something that you expect to be a controversial point, realising that the likely result is going to be a certain number of responses implying that holding that opinion is stupid is a pretty strong incentive to avoid posting in the first place. Suppressing the expression of thoughts isn't beneficial, even if you're not doing so deliberately.

Now, the obvious response to this is to accuse the CoC of enforcing some sort of Tellytubby land where people are nice to each other and it's all a mutual love-in. Which isn't, in fact, the case. Disagreeing with people is fine, but disagreeing in a factual and precise manner is significantly more useful than disagreeing in a manner that gives the strong impression that it would have been more helpful if the original poster had died rather than sending the mail in the first place. You're not even obliged to behave in a friendly way towards people - just avoid behaving in a clearly unfriendly way. A good rule of thumb is probably that if you think it would potentially result in a complaint to your boss if you did it in a moderately large company, then doing it on an Ubuntu mailing list is a bad plan.

So, the original point. The joke itself implies that there's no way to know why women behave the way they do, even to the point where a (presumably) omniscient god has no idea. Unsurprisingly, telling a large body of potential contributors that their behaviour is irrational[2] isn't a great way making them feel like their contributions are welcomed. But, oh no! It's actually ok - we're supposed to replace "women" with "newbs", because suggesting that basically your entire pool of potential contributors that their behaviour is irrational is much less bad.

Cough.

We then get to Caroline's post, an expression of frustration about an aspect of life. It contains the word "Fuck", which doesn't seem unreasonable in context. There's then a complaint from someone claiming that it's offensive, which means we have to consider a few things. Firstly, is the author of the complaint actually offended, or just attempting to claim that there's hypocrisy involved? Given the context, I'd suspect the latter. Arguably this isn't really the point, though - the issue isn't whether a given person is offended, it's whether it'll offend someone who would otherwise potentially make a valuable member of the community. Right now, I'd suspect that we (as a community) don't really object to it strongly. Google suggests that swearing isn't unheard of on the lists. While swearing in the context of a personal attack is pretty clearly a violation of the CoC[3], people don't seem to be objecting to it in general.

So, is it unacceptable to swear within the Ubuntu community? Right now, I'd say that (in appropriate contexts) it's not a problem. If a moderate number of people claim that the current degree of swearing dissuades them from being involved, then it may become a problem. Accusing someone of hypocrisy for swearing (not clearly a CoC issue, especially in context) after they claimed that a sexist joke might be, well, offensive? Yeah, I'd say that that one's pretty clearly unacceptable. Telling people that it's pretty funny and that they're wrong to be offended? History suggests that it's not a good plan. If you think someone's wrong to be offended, then a good first step is trying to figure out why rather than just dismissing them.

Of course, there's also the entirely secondary issue of what the roles of planet sites in the community are. Personally, I like to think of them as opportunities to see what people are really like rather than being reflections of people's behaviours on project mailing lists[4]. From that point of view, I think that the bounds of acceptable behaviour should be a little more liberal. If people disagree, that's not a problem and things can change - but expecting people to change their behaviour before there's any sort of general opinion that it's unacceptable isn't ok.

Which leaves open the question of whether posting sexist jokes to mailing lists is on the acceptable side of the CoC[5] or not. I'd be surprised if the ubuntu-uk community genuinely thought so, but if they do then there are pretty clearly issues that we need to solve. As far as I'm concerned, justifying that sort of behaviour has no place in Ubuntu.

[1] Yes, I'm aware that there's a certain degree of irony there
[2] For the record, yes, I do believe that women often behave in irrational ways. So do men.
[3] I find it surprisingly hard not to write that as COCK. Too much Jerkcity
[4] Which is, to an extent, why this journal isn't on Planet Ubuntu - there'd be potential for things I said to be mistaken for some sort of official Ubuntu position, which would end up limiting what I could say, which would kind of ruin the point
[5] Must... resist... temptation...

Matt Brown: My DPL Vote


- - -=-=-=-=-=- Don't Delete Anything Between These Lines =-=-=-=-=-=-=-=-
e0acebd2-71f1-4df8-ae4d-50355ad7aa81
[ 6 ] Choice 1: Wouter Verhelst
[ 6 ] Choice 2: Aigars Mahinovs
[ 3 ] Choice 3: Gustavo Franco
[ 3 ] Choice 4: Sam Hocevar
[ 2 ] Choice 5: Steve McIntyre
[ 4 ] Choice 6: Rapha l Hertzog
[ 1 ] Choice 7: Anthony Towns
[ 6 ] Choice 8: Simon Richter
[ 5 ] Choice 9: None Of The Above
- - -=-=-=-=-=- Don't Delete Anything Between These Lines =-=-=-=-=-=-=-=-
My rationale follows, if you care... I've been pondering how to vote for well over a week, and I'm still not entirely happy that the ballot I've chosen accurately expresses my wishes, but it's the best approximation of them that I can come up with at this time. My dilemma has two parts
  1. I don’t think the current office of the DPL is effective, due to how it is viewed by a significant proportion of the project (if not an outright majority), so regardless of how much I might like the goals and ideas proposed by an individual candidate I’m very pessimistic that being electing as DPL will actually bring those things to pass.
  2. Much of what will make a good DPL is about how a specific set of ideas and goals will be put into action, and how interpersonal relationships between the DPL and various parties will be managed. This comes down to the personal character, experience and leadership skills of the candidate While I can form some level of an opinion about each of these aspects from mailing list archives, and IRC, etc. I don’t really feel comfortable making judgements in these areas until I’ve actually met them in real life. Too many people come across badly in the (severely limited) online communication methods we use, and are actually very decent reasonable people in real life. Out of the candidates, AJ is the only one that I’ve ever actually met and talked to.
Given these problems, my first step in choosing how to vote was to eliminate those candidates who I will rank below NOTA based on their published platforms. I have nothing personal against any of these three people, but given the other possible candidates I think that we’d be better off having another election than just electing one of these three for the sake of it. That leaves five remaining candidates: *yawn* It’s made me tired writing all of that out. Maybe sometime soon I should explain in more detail exactly what parts of Debian’s governance model I think need changing or maybe I need to wait until I’ve achieved a few more technical things and gained enough respect before anyone will listen to my opinions…

Nico Golde: Aaaaarrrrgghhh

Today I wanted to learn for my examns and thought I'd make a quick dist-upgrade before, looked at the packages, found nothing important to make a backup and DANG

All installations with debconf usage didn't succeed like here for example:

dpkg: error processing samba-common (--configure):
subprocess post-installation script returned error exit status 1
dpkg: dependency problems prevent configuration of smbclient:
smbclient depends on samba-common (= 3.0.24-6); however:
Package samba-common is not configured yet.
dpkg: error processing smbclient (--configure):
dependency problems - leaving unconfigured

I straced, asked in #debian-devel, read source code and really didn't find the problem.
Thanks Mika for noticing the same problem before I did and the fix was sooo easy.
ln -sf /bin/bash /bin/sh

Never ever link /bin/sh to zsh on Debian :-)

Alexander Wirt: LinuxTag2007 Debianday

The Linuxtag Orga offered us a Debian track so I asked for submissions, but for now we only have 3 submissions. This won't be enough for a track, so this is my last call for papers for the Debian Track. If you plan to attend to the Linuxtag and want to talk about Debian, use the VCC to submit a talk there. Additionally my request for help had nearly 0 feedback. If you want to have a Debian Booth at the Linuxtag its now the time to get in touch with me...

Enrico Zini: remoteform-weird-error

TurboGears RemoteForm tip In case your RemoteForm misteriously behaves like a normal HTTP form, refreshing the page on submit, and the only hint that there's something wrong is this bit in the Iceweasel's error console:
Errore: uncaught exception: [Exception... "Component returned failure
code: 0x80070057 (NS_ERROR_ILLEGAL_VALUE) [nsIXMLHttpRequest.open]"
nsresult: "0x80070057 (NS_ERROR_ILLEGAL_VALUE)"  location: "JS frame ::
javascript: eval(__firebugTemp__); :: anonymous :: line 1"  data: no]
the problem can just be a missing action= attribute to the form. I found out after:
  1. reading the TurboGears remoteform wiki: "For some reason, the RemoteForm is acting like a regular html form, serving up a new page instead of performing the replacements we're looking for. I'll update this page as soon as I figure out why this is happening."
  2. finding this page on Google and meditating for a while while staring at it. I don't speak German, but often enough I manage to solve problems after meditating over Google results in all sorts of languages unknown or unreadable to me. I will call this practice Webomancy.

Next.

Previous.