Search Results: "ag"

18 January 2007

Gunnar Wolf: Then fix the software, not the format!

Erich: Of course that I know XML's obstination on manually matched closing tags is intentional - That and several other points simplifying SGML were what in the end allowed SGML to succeed (and wildly). However, I still not agree. If you have a tool that generates broken files (which are expected to be read, no less, by any other arbitrary tool which does not require to be very bright), whose fault is it? The file format? No - The generator's. Besides, if you have a configuration format that's expected to be used by al kinds of tools, and sent over every conceivable type of configuration channel (including slow and expensive ones, such as mobile phones), or that can contain arbitrarily deeply nested structures and become just huge, shouldn't your priority be to make the protocol less repetitive instead of more? If you want a format to be robust, yes, you should insist on well-balancedness (what's the last time you were able to compile C code with unmatched braces?), and reject unbalanced documents (possibly even, yes, pointing out where the match was probably broken - Yes, this last point favors XML over simple braces, but still, a compiler often makes a decent job at finding where a nesting problem lies)... Yes, I don't have the authority to question what has become a world standard - But I still have to be persuaded XML is the way to go for many (no, of course, not all) of its uses.
No, I'm not pushing YAML too hard - I just happened to like it for several uses, but I'm far from an informed fan. I'm just bashing XML, which is fun enough for me ;-)
Oh, and about your blog: Right, it seems the culprit is Google Reader. Planet links just fine, and using RSS 1.0, the only link I get is the right one. Planet's RSS 2.0 generator should be to blame then.

MJ Ray: Is this storm coming for you?

One of the bad things about moving to live in the west country is that we are now on the front line of any storms that leave the Atlantic for England. To make matters worse, we are on a hillside and some idiot put the boiler vent facing into the prevailing wind, so I have to relight it at least once each bad storm. Today's storm is the worst yet. Average wind speed is reported by TV as 35mph, gusting over 70. Anything that wasn't bolted down or very very heavy and hadn't already been blown away has now gone. I keep going out quickly to do things like add weights to the bin or wedge one of the garden tables against the fence, but I don't feel like fixing my bicycle and riding to town along the coast today. On the first trip out, as I was checking the roof for damage, one of the neighbours' six-foot fence panels flew off. On the second trip... well, this used to be a fence on the other side of the path by our back wall: Photo of no fence Now it's just a concrete post surrounded by debris. If this is part of the storm heading for Erich Schubert then I think Munich has done the right thing.

Thijs Kinkhorst: On government crisis response

I'm here at university centre de Uithof. The government is advising against any travel due to a 10-11 Bft storm passing over the country. Earlier today, a tall crane collapsed onto one of the university buildings here, leaving luckily only a couple of people wounded. Since there's food and heating here, I'll just stay for now. The Ministery of Interior is boasting about its new initiative: a crisis website meant for situations where normal government websites cannot keep up with the high load: www.crisis.nl. According to the ministery, this website can handle "10 milion of visits per 5 minutes". The website performs very badly, being more down than up. A closer inspection seems that the site is using some content management system based on cold fusion. The website is also illustrated with nice pictures. I wonder where the ministery has sourced its advice. What about serving simple, static HTML pages (perhaps generated from some CMS)? How about not creating a lot more load with superfluous illustrations? Just a few ideas.

Clint Adams: Things that did not happen at Google last night

I was en route to an event by myself, when I received an invitation to accompany some people to a movie. Rather than simply declining, I asserted that there was no way in hell I'd cancel my plans to go see some inferior film with that particular crowd. So, what did they do? They cancelled their plans and followed me. Now one of them was apparently in heat, for instead of trying to enjoy the event, she decided that it was some kind of speed dating event and before long she was coveting one of the event staffers. This offends me on so many levels, I won't even bother to explain. Eventually she managed to discourage herself so much that she went home, but the sanctity of the event was still spoiled. Some random person came up to me and asked me to explain dynamic DNS to him. Do I look like the type of person who would have and give the answers to such questions? I opted to not beat him down with a chair or steal his Treo and slit his throat with it. Instead, I sat down and began explaining. Now some girl, who had marked me as her prey, spotted me sitting there next to this guy who bore a resemblance to Andrew Suffield. Seeing his long, flowing hair, she immediately concluded that he was a girl, and therefore a threat to our marriage or whatever she had concocted for us in her dreamworld. So she swooped down out of the sky, ready to claw his eyes out, realized that he was probably male, and breathed a sigh of relief. I concluded my exposition, and the young lad rewarded me for my daily good deed with a thank-you and a shake from his slimy, sweaty hand. Ew. Ew. Ew. EW! He departed, but the girl did not. She had with her about seven wingmen. I am told that they are wingirls (and not winggirls) because most of them are female. I am not sure that I agree, so I will call them wingnuts as a compromise. Now I have to wonder why one would need seven wingnuts. It seems to me that if one is planning ahead to require the assistance of seven, one must be hunting prey that will pay a high cost to escape. Furthermore, if one is fixin' to (or fi't'n't'a in Alabama) ensnare and entrap such an unwilling quarry, what are one's motives? What is the underlying scheme? The possibilities are horrifying. It was difficult to have a good time, but I did it anyway.

Erich Schubert: A major storm is heading our way, according to weather forecasts

This is the first time I can remember, that the university is recommending to not hold lectures and go home early. They are expecting the ministerials to formally close the university, like they did with schools at noon. This will be the strongest storm we've had in years, apparently (well, others have been similarly strong I think, but not as big). It's now 15:18, and when I look out of the window it's still surprisingly quiet. The trees are moving, but only slightly. But it's the wind protected side of the building. The university warning email said that they expect public transport - and Munich has a huge public transport system - to stop service at 18:00 "the latest". Guess I won't be able to go dancing tonight then. It would start at 19:00, and I need about 30-40 Minutes by subway to go there. I guess they will shut down subway to make sure noone gets stuck in the tunnels if there is a power loss. Such huge storms are a threat to power supply. [Update: 21:08 you can tell now it's a storm, but it's not bad here. But all train traffic is stopped in bavaria, grounding a few hundred people who didn't make it home in time. They've been taken care of. In Munich, busses and the subway are supposedly still in service. I went to go dancing, but there was a note on the door it's closed tonight; same for the karate training of a friend. From what I heard in the radio there were only few fatalities so far and limited damage, but we havn't reached the peak of the hurricane yet. Few reports of blackouts in some small towns. Schools remain closed tomorrow.] [Update: 19. Jan, noon - trains apparently still aren't back to schedule everywhere, but the system is recovering. Not many major damages reported in the media and not many fatalities. It's still windy. IMHO it was "typically german" how we handled the hurricane, urging people to stay at home and not risk much...]

Miriam Ruiz: Experiments with usplash themes

I’ve been doing some usplash-related stuff these days, but I’ve come across some problems. First of all, I want to be able to track some hardware information from within the experimental theme I’m working on, and that means reading some stuff from /proc and /etc (and maybe from /sys in the future, I don’t know yet) if it is available. It seems to work OK if I open the file handler from the beginning and keep it open, rewinding to the start of the file when I need to read its contents but never closing it until the end of execution. If I instead open the file, read from it and close it again, it seems to work properly for the beginning, but after a while it stops retrieving data. I guess that when the real root is mounted i lose the capacity to interact with the filesystem from my code, like opening new files. The second problem is somehow related. I want to be able to save a log file of the execution of the program, but I don’t seem to be able to open a file for writing, at any moment in the execution of the program. I guess it’s the same problem, accessing the filesystem from a program that is started from initrd.img instead of from the real disk. Is there any way to connect the program with gdb during real execution? I seem to be getting SEGFAULT errors with some things I try, but it’s hard to debug without any data (not even logs, as I cannot open a file for writing at any moment). Maybe being able to dump a core when a SIGSEGV signal is received would do. More on this, in future posts.

MJ Ray: sxw2text en python

I wrote sxw2text as a 5-minute email-helper hack some years ago. Now I notice that there's also a version of sxw2txt in python on a French site, but it doesn't seem to do any checking at all, not even the minimal bit of my script. Update: following more links, I found odt2txt.py which outputs markdown and is probably a better choice if you have python installed.

Biella Coleman: 5 year moving wall

Despite my claim that I would post the Science articles on money once I got access via the U of A library, well, they have a 5 year moving wall, so I won’t be doing that anytime soon. If anyone does have access to the article because they are subscribed to the magazine or are a member of AASS, please feel free to pass along copies to me. My contact info can be found on my CV.

Brett Parker: So, another day...

We're at 5 minutes to midnight, and it appears that Joost is the new TV - it appears to be getting some reaonable coverage at the moment, but looks far too much like another closed app using proprietary protocols - the world has gone DRM crazy - someone stop it, I want to get off. DRM is a myth, if you can play it, you can copy it (not neccessarily easily, but it's still possible). The NewScientist article is quite interesting in the claim of "piracy-proof"... if you can play the video, you can capture it, if you can listen to music you can record it, when is the media industry going to realise that DRM is a very dumb idea that shouldn't have been allowed to get this far - pirates can (and always will) find a way, how about going back to trusting your users rather than tarring them all with the "you're a pirate" brush. The Register 's article appears to have a typo in it... well, unless you really do need a 500mbps line in order to use Joost - seems unlikely, I think they meant 500kbps. Anyways - nice idea, not sure that it's better than anything else though, I'll continue buying DVDs that let me play back when I have no interweb connection. Back to being at 5 minutes to midnight, eeep! We're all going to die! Maybe it's the introduction of DRM that's led to so much terrorism, or maybe just that the media industry appears to be out to shaft everyone. Bah. Only one other article has caught my attention this morning, it's another look at the iPhone, it appears that no one is actually getting a reasonable amount of time to play and review this device, and it brings up a number of my issues with the iPhone. So, yes, that's the morning covered - maybe more this evening.

Guido Trotter: Real Life Race Condition

When one studies Race Conditions in Operating Systems they tend to seem some weird computer problem, maybe even a hard one to understand… Yesterday I had the proof they’re actually a real life issue! Our office in zurich has a form to order dinner online, to be filled in before 5pm, which I did, just after my colleague, one minute before 5 or so. The system got my order and accepted it. It also sent me mails to remember to fetch the food except… there was none! It seems like my pizza got caught in a race condition between the actual ordering and the official closing of the web form. (My collegue’s food was delivered normally ) Race conditions seem to be much more real when what gets lost is your dinner! (Well, I managed to eat anyway, later, luckily!)

Russell Coker: new release of postal

Today I have released a significant new version of my mail server benchmark Postal! The list of changes is below:
Thanks to Inumbers for sponsoring the development of Postal.

I presented a paper on mail server performance at OSDC 2006 that was based on the now-released version of Postal.

I've been replying to a number of email messages in my Postal backlog, some dating back to 2001. Some of the people had changed email address during that time so I'll answer their questions in my blog instead.

/usr/include/openssl/kssl.h:72:18: krb5.h: No such file or directory
In file included from /usr/include/openssl/ssl.h:179

One problem reported by a couple of people is having the above error when compiling on an older Red Hat release (RHL or RHEL3). Running ./configure --disable-ssl should work-around that problem (at the cost
of losing SSL support). As RHEL3 is still in support I plan to fix this bug eventually.

There was a question about how to get detailed information on what Postal does. All three programs support the options -z and -Z to log details of what they do. This isn't convenient if you only want a small portion of the data but can be used to obtain any information you desire.

One user reported that options such as -p5 were not accepted. Apparently their system had a broken implementation of the getopt(3) library call used to parse command-line parameters.

Rodrigo Gallardo: Sexism

I have a pretty thin voice for a male. Thus, it’s not rare that people will take me for a female when on the phone. I’ve rather gotten used to it, and it no longer bothers me that much, specially since it usually happens with people I’ll never speak to again anyways (such as telemarketers). But, today, this guy called asking for info on an apartment we’re selling. I gave him the price, he said something about calling back later, called me “Ma’am”. So far, business as usual. He then reiterated he’d call back and ended by saying something like “thanks, cutie”! He actually hit on me! And it’s just so upseting! I mean, what could posibly make him think I want total strangers to call me that? So now, I just try to think about how it must be to be a woman, and have random strangers hitting on you all the time and on top of that thinking you’ll like it, and I think I just could not bear it. Most of us males really are just pigs. Ugh.

Biella Coleman: Science on the Money Effect

Within 36 hours of my return to Edmonton, from the blowing Caribbean winds to the still calm of white snow, I have fallen sick with a cold, that while not a flu (at least not 24 hrs later), is still a severe cold, knocking on flu’s door. But ever since I had various horrible experiences having to work with horrible colds (like during my qualifying exams), I don’t mind colds so long as I can stay at home and let the cold run its course, which at least is my current predicament. Because I have been parked at home, I have spent a fair amount of time on the computer today playing catching up with blog entries and emails and I came across a few potentially interesting articles about the effects of money on tieguy’s blog (run law student who knows a heck of a lot about tech, law, and free software but his blog seems to be down at the moment). He linked to pair of articles on the psychological effects of money. I have not read the articles yet (and will post them, hopefully tomorrow once I get access to them via my U of A account), but I bet this will be of interest to some Debianista’s given the recent debates and controversies over the injection of money into Debian via the Dunc-Tanc project. Now a disclaimer: I don’t have a position on Dunc-Tanc, and this is so for many reasons–the primary being I have not delved deeply into the issues an all of the debate and discussion and well, I also experience the ethical issues somewhat more of an outsider, though I do want to see Debian survive well into the future. So I have keep mostly mum on the topic but later I may have more to say. I also have yet to read the articles and am usually a little suspect of psychological experiments that purport to have universal applicability (and am not sure if these fall into this class) so I am not sure how relevant these will be to this particular case. But nonetheless, here they are, and hopefully someone will find some use in them (and sorry if they have been posted here, I am very behind on planet, thanks to dial-up for a month). On a somewhat related though different note, check out Joseph Reagle’s excellent summary of how online communities work well. Now time for much needed sleep.

Christine Spang: SIPB IAP Debian Class Slides

One of the student groups I’m involved in at MIT is the SIPB, the Student Information Processing Board. It’s a volunteer computer group so old that they called it “information processing” back then. They run a bunch of different servers and provide lots of cool services for the MIT community. One of the things that they do is teach classes about various things related to computing during IAP, MIT’s Independent Activities Period. So back when I had just gotten started with the group, one of the members was like, “Hey! You want to teach an IAP class!” and, being a frosh, I was like, “Er, I didn’t know that… but, um, okay.” So I taught one about Debian, of course. This year’s topic was Giving Back: Contributing to Debian and the New Maintainer Process. It was a three-day series, as described here. The lengths of the talks varied, with the first day being ~45 minutes, the second being ~1.5 hours, and the last being ~30 minutes. All of the lecture materials can be found here. The audience was on the high side of technical, and a lot of the slides were more to nudge me into what I was talking about rather than to be really awesome slides. Which probably should have involved less text. The packaging crash course definitely drew a much larger audience than the first or third parts, which mostly consisted of people I know who work on Debian-Athena, the unexpected appearance of my friend Aaron Swartz who’s in town for a couple weeks, and a few faces that I didn’t know. Maybe next year I’ll try to plan something that just involves a more intensive packaging course. Because, although I procrastinated, pushed the original lecture dates back a week from original plan, and made my lecture slides on the day they were presented, I actually did have fun. And it’s good practice.

Erich Schubert: More on XML

Gunnar, first of all, let me point out that repeating the closing tag makes the format more robust. If the parser encounters a non-matching closing tag it knows that something isn't right. Thats a feature, not a bug. IIRC SGML had shortened closing tags (</>) and closing tags were partially optional; if the parser encountered a closing tag it would assume all other tags closed up to the previous matching tag. I guess some people who used SGML and were involved with XML had made the experience that this was a bad idea. It makes parsing harder and the file format more likely to break. If XML would only be written by perfect tools, we wouldn't need the verbose closing tags. We could probably use some binary syntax like "\1tagname\0\2contents\0". Parsing would be fast and easy. Or something along
tag   "string contents" tag    
(which I like a lot). Or S-Expressions. But it is a strength of XML that you can detect certain common errors such as incorrectly nested tags: XML data is generated by broken tools all the time, so we need a syntax which allows us to detect such errors. As for my RSS feed: the links are correct. The tool you are using mishandles the guid tag, which is just a unique identifier; the sole responsibility of my blog is to make them unique. Many tools use a link for the guid, but that is not required by the spec. In fact my blog is quite verbose about not using the full URL there: isPermaLink="false". Instead it provides the "link" element. Look at the source of my RSS, no repeated "en/" in there. Planet is broken here, and your reader:

Evan Prodromou: 27 Niv se CCXV

The snow we got a couple of days ago hasn't abated here in Montreal. We had a cold snap right after the snowfall, so temperatures have been around -20C for the last couple of days. Last night Maj had her French classes at the Montreal YMCA, and Micaela had spent the day seeing art museums and visiting bookstores downtown. (She's a poetry nut -- loves finding books by Canadian poets, francophone and anglophone, that she can't get at home in Melbourne.) So Amita and I went to meet her about halfway from our local Metro station, at the Pizzad lic on Ave Mont Royal. I covered Amita in about 10 layers of clothing, then put her on her little purple sled and dragged her up towards Mont Royal. She kept wanting to get up and walk herself ("Up!" is her newest and most insistent word), but when she was standing she wouldn't walk at all. So it was a long time getting to the restaurant. When we got there, Micaela was leaning against the front door shivering. Apparently Pizzad lic is out of business (bummer!). A graver cause for concern, however, was Amita's face: the parts showing around her scarf were tomato red from the cold. Gah! We rushed her home (crying all the way), and when I got her undressed she had a red stripe across the middle of her face like the people in those frozen pizza ads. I bathed her cheeks in cold water, and then gave her a hot bath, and by the time she was out she was calm and regular-colored. We got pizza delivered at home, which she enjoyed a lot. No permanent damage, as far as I can tell, except to my self-confidence as a competent parent. When Maj got home later that night, she reminded me that for extremely cold weather you have to put Vaseline on babies' cheeks to protect their delicate skin from the cold. Amita and I went to Loblaw's this morning to do some grocery shopping (in the car), which was a much more reasonable experience. Tonight for dinner I made pat chinois, a plate with baked mashed potatoes and ground beef known to English-speakers as shepherd's pie. (I make it with mushrooms and veggie beef.) Here in Quebec it's called "Chinese casserole", for unknown reasons. The best guess is that French Canadian lumberjacks worked at sites in the USA where it was common to have Chinese immigrant cooks. Confronted with a potato dish they didn't recognize, they assumed it was a meal from the cooks' home country. I don't know if this origin is true or not, but it sure is a good story. Anyways, it's good stick-to-your-ribs food, which you need on a night like this. When it gets down to -28C (as it will tonight), you need a lot of carbs running your internal furnace for you. tags:

Piknic lectronik The good snow we're having means that one of my favourite events of the year is going to be white and cold and fluffy and fun. For the last few years, Piknic lectronik, the group that produces weekend outdoor parties on le Ste H l ne on the site of Expo 67, has done a single, crazy outdoor party in the dead of winter. This year, the party, called "Igloofest", is happening in the Vieux Port on the Quai Jacques Cartier. It's going to be two nights, January 19-20 (that's 29-30 Niv se for you aficionados of the French Revolutionary Calendar), from 5 to 11PM. We're going out on Friday night -- our pals from Ninjatune are part of the party, and our friend DJ Ghost Beard is on that evening. It should be a great time; I'm really looking forward to dancing in the snow and having vin chaud afterwards. tags:

BarCampMontreal II: Camp Harder Just got the news from my blogroll: Sylvain Carle notes that the second BarCampMontreal will take place April 28th. That's going to be close for us; we'll be at the Wikitravel Get-together in Puerto Rico up until the 22nd. And we plan to fly there from Miami. But, hey: stranger things have happened. Our BarCamp last year was an absolute blast (see my follow-up on Journal/1 Brumaire CCXV); lots of fun people doing interesting things in fun ways, and then talking about it. I'd really like to be there for this next one. tags:

Louisianne I'm happy to see that in an alternative history universe called Ill Bethisad, the French Revolutionary Calendar is used in the North American francophone nation of Louisianne. On the downside, my city of Ville-Marie is in the royalist enclave of New Francy, which uses the boring old calendar. Ill Bethisad is an impressive alternative-history project, and the wiki format seems to really be conducive to its further development. Great idea, and nice to see the ol' FRC included. tags:

Baby Lessig Congrats to Lawrence Lessig and family on the birth of a new son. tags:

Dirk Eddelbuettel: New ip

In the middle of last night, Comcast once again decided to move 'us' (and presumably gazillion others) to a new IP address. In the process we moved from the 24.14.* subnet back to 67.174.*.*.* subnet. For once, I didn't seem to have blogged about the last chance but it feels like the previous assignment had held up forever, i.e. a couple of months. Thanks to the good and quick dns service at Gandi, my two domains eddelbuettel.com and quantian.org were again visible within minutes of me finally noticing the chance this morning. Sorry for any service interruptions in the meantime.

Erich Schubert: Putting a CV on the homepage

Dear Lazyweb, what is you opinion to having a CV on your homepage? There are definitely privacy issues associated with this. Even having your birthday there makes abuse easier. On the other hand, many people will write a blog posting on their brithday and thus expose it anyway. I'd never put a full CV on the homepage (a good CV is adopted to the audience anyway), just with the bare minimum such as the university graduation and such. Having some project and research information there might get you some interesting job offers; putting a mini CV there as "teaser" along with the offer of a full CV on request could prove beneficial. On the other hand, such as CV could be a dangerous source material for "social engineering [wikipedia]" attacks. The maiden name of your mother used to be a suggested backup password for many websites, and is found in some CVs; people tend to trust people more that claim to have been on the same school or university (even if they can't remember them; especially if they are made feel embarassed by not remembering...). Where is your privacy limit? How much of your privacy did you expose on the web already anyway? Do you still avoid having a one-stop download of that data? Do you think there is a way back, with all the indexing, caching and archiving of web pages? My blog does intentionally not have comments; to reply please post in your own blog or email me at erich@debian.org; I read a couple of OSS planets and I'll also notice if you link to this blog entry and ping technorati.

Erich Schubert: Putting a CV on the homepage

Dear Lazyweb, what is you opinion to having a CV on your homepage? There are definitely privacy issues associated with this. Even having your birthday there makes abuse easier. On the other hand, many people will write a blog posting on their brithday and thus expose it anyway. I'd never put a full CV on the homepage (a good CV is adopted to the audience anyway), just with the bare minimum such as the university graduation and such. Having some project and research information there might get you some interesting job offers; putting a mini CV there as "teaser" along with the offer of a full CV on request could prove beneficial. On the other hand, such as CV could be a dangerous source material for "social engineering [wikipedia]" attacks. The maiden name of your mother used to be a suggested backup password for many websites, and is found in some CVs; people tend to trust people more that claim to have been on the same school or university (even if they can't remember them; especially if they are made feel embarassed by not remembering...). Where is your privacy limit? How much of your privacy did you expose on the web already anyway? Do you still avoid having a one-stop download of that data? Do you think there is a way back, with all the indexing, caching and archiving of web pages? My blog does intentionally not have comments; to reply please post in your own blog or email me at erich@debian.org; I read a couple of OSS planets and I'll also notice if you link to this blog entry and ping technorati. [Update: first interesting replies coming in - let me quote:
If I do get a call from someone who spotted it (rare, once every 3 months), the recruiter is 99/100 an asshole he isn't going to ask you to do something you want. Really!
Yeah, I can imagine that. One of the reasons I've put up a mini CV is that every few months I'm asked some of these basics usually found on CVs because of my art - I have some poetry on my page that is very popular, and every now and then someone wants to use it in school; teachers always want people to be able to tell something about the author. In such cases, I find it actually worth to highlight that I'm a math and computer science guy, to make people see that it isn't an either-or thing, language and science.]

17 January 2007

Norbert Tretkowski: Broken newsreaders

If you upgraded to slrn 0.9.8.1pl1-22 or newer and now realize that you see question marks instead diacritical chars (e.g. german umlauts) in the subject and/or in the authors name when looking at the article overview, please read #406210 before filing yet another bugreport. This is not a bug in slrn, but a bug in the newsreader which was used to write the corresponding article. Update: 0.9.8.1pl1-25 contains a workaround for this problem, see NEWS.Debian.gz for details.

Next.

Previous.