— Well, what exactly do you do?

— I make computers work.

— (?) …How do you mean?

— You have used computers, right?

— Yes.

— They often don’t work, right?

— …Yes.

— Now try to picture what happens when you have about a thousand of those.

— …Oh.

Even when working at Layer 2, ExtremeWare takes advantage of information available at other layers to establish and enforce QoS and access control policies. Layer-independence means that when an Extreme Networks product is routing or switching packets, QoS and access control decisions can still be made using criteria pulled from Layers 1–4 or from other sources.

(Source: «Extreme Networks Technical Brief: ExtremeWare Operating System»)

This problem is even more annoying because wl is an Emacs application, and the Emacs OS lacks multitasking (they tell me that’s because it lacks closures, and that is because its flavour of Lisp uses dynamic scoping, and that’s hard to change). So your whole OS stays blocked while wl filter your messages.

This same issue killed any hope of a pure-Emacs web browser (like w3). People have circumvented it by calling external, non-Emacs applications, which can run asynchronously thanks to the host OS (like w3m-el).

You can do the same to organize IMAP mailboxes with this nifty little program called imapfilter. It uses Lua for configuration and knows how to filter efficiently, by asking the server to match patterns and move messages. It only deals with message IDs. The thing is so fast, it’s usually done almost instantaneously.

Here’s how to hook up imapfilter to wl:

  ;; on your wl configuration file, or wherever you keep this stuff
  (when (and (file-readable-p "~/.imapfilter/config.lua")
             (executable-find "imapfilter"))
    (eval-after-load 'wl
      '(add-to-list 'wl-folder-check-entity-pre-hook
                    (lambda ()
                      (message "calling imapfilter…")
                      (if (eq (call-process "imapfilter") 0)
                          (message "imapfilter ran fine.")
                        (message "error running imapfilter!"))))))

I use the when condition for robustness, because I copy my emacs conffiles to several machines. The eval-after-load prevents startup delay. The meat of it is the add-to-list hook, which will cause imapfilter to be called whenever you synchronize folders (key s on folder view).

Condescension

A typical software example is “The System can frobnicate, twiddle, and twirl, but don’t worry, you can do it all very easily with the new System Icons!” When you say “we made it easy” what you really mean is “We think you users are stupid, so in Our infinite wisdom we graced thee with operations so simple even a caveman would understand”. Another example of condescension are most uses of the words “trivial”, “obvious” and “clearly” in math texts.

There is such a thing as natural complexity, but bad writing and bad design are much more common. Don’t blame the readers for your bad writing, and don’t blame the users for your bad design. If you can afford, hire a real editor and a real UI designer to fix them. If you can’t, well, then don’t blame your clients!

Self-congratulation

“The System was brought to you by the Department of Systemic Systematization, under the kind patronage of the University of Systems Research. It is provided free of charge to the public in the USR’s mission to help education as part of our social responsibility program that” etc. etc. etc. Self-congratulation is not offensive like condescension, but it’s still annoying and useless. It violates the first commandment of technical writing: Thou Shall Not Waste Words. The reader opened your manual to learn how to use the system, not to learn how amazing you are.

Often self-congratulatory sections are written against the will of the author, due to some kind of political pressure. If you’re a writer in this situation, see if you can tuck it politely out of the way, in the last section or as small print.

“Ah, but you see, friend, Appendix C sucks. Verily I say, if thou sendest them XHTMLs as text/html, then they’ll be tag-souped, so there’s no point. Thou might as well write the plainest HTML.”

Word. And further, I hate composing Appendix C code! I hate having to duplicate id and name and lang and xml:lang and I hate those lame spaces before slashes. It just defeats the purpose of choosing XHTML as the shorter format.

Way back then, when I still thought XML was a good idea, I decided to solve this with content negotiation and XSL. I wrote this stylesheet to convert XHTML 1.1 to HTML 4.01.

sabcmd xhtml2html.xsl eatsquaredonuts.xhtml eatsquaredonuts.html

With some help from make(1), I can write pretty, succinct, non-appendixized XHTML 1.1 and automatically create HTML from it.

Now what? Well, I gotta tell Apache to send XHTML to browsers that support it, and HTML to others. A first attempt could go

Options +MultiViews
AddType application/xhtml+xml;charset=UTF-8 .xhtml
AddType text/html;charset=UTF-8 .html

Then, instead of making links to /eatsquaredonuts.html or .xhtml, you simply use the URI /eatsquaredonuts. Each request, Apache will have a seat with the browser and they’ll decide politely what version (“representation”) to send.

Looks great, but when I try a stock, run-of-the-mill Firefox, I get the HTML version. That’s not so bad, but you know, the XHTML version is sexier. I’m using this for my resume, for devil’s sake; how am I supposed to impress the guys at Canonical with boring old HTML? I need a job, Firefox. Just what are you asking Apache in those private conversations, don Firefox?

Accept: text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8

Ah, see, Firefox say it will accept HTML and XHTML equally well. If neither is available, it will take application/xml, and if even that isn’t in the menu it will make do with anything (*/*). The order is controlled by the q parameter; it’s an implied 1 if omitted, so Firefox is leaving the choice to Apache, and Apache chooses HTML (perhaps because Firefox mentioned it earlier —I guess old Apache is a lazy man).

Is there a way to tell Apache to prefer XHTML in case of ties? It’s not clear from the docs, but turns out there is:

AddType application/xhtml+xml;charset=UTF-8;qs=1 .xhtml
AddType text/html;charset=UTF-8;qs=0.99 .html

Now Firefox does get my beautiful, precious XHTML, and non-XHTML browsers shouldn’t even see it (since it won’t be in their Accept headers). You know just who am I talking about, right? Let’s test:

Internet Explorer can’t download blahblahblah.

Internet Explorer could not open this site blahblablah not available or could not be found blahblahblah.

wat.

I get it, this message means HUURRR IM A BROWZZARR WHATS THIS XML THINGIE HUUURRR. But why is Apache sending XHTML in the first place? What is the browser asking Apache?

Accept: image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, application/x-shockwave-flash, */*

* * *

There’s no text/html.

This is a web browser, and there is no text/html.

It relies in the order to decide content instead of q parameters, and has a */* at the end, and there is no text/html.

* * *

MSIE, the Bane of the Internet. MSIE, the Breaker of Standards. MSIE, the Blight of Developers.

* * *

Because of MSIE, even our first attempt isn’t safe. We’re forced to tell Apache to resolve ties in HTML’s favor:

AddType application/xhtml+xml;charset=UTF-8;qs=0.99 .xhtml
AddType text/html;charset=UTF-8;qs=1 .html

As far as I can tell, the only way to get Firefox to see XHTML without choking MSIE would be some very convoluted new configuration option to Apache. I’d have to tell Apache, “if text/html and application/xhtml+xml tie, then choose XHTML, except if they’re tying in a */* because the browser didn’t even ask for HTML, in which case you should send HTML”. I’m not sure I even want to ask this feature. I wouldn’t know what to call it.

* * *

The solution, I guess, will be to wait for HTML 5 to turn this dispute academic. See you in 2022…

This is a simple wrapper over exiftool (debian/ubuntu: in package libimage-exiftool-perl) to make it easy to set EXIF geographic information (GPS tags). It is able to understand the decimal format for latitude/longitude that you can get from google maps, wikimapia etc.

With exiftool:

  exiftool \
      -GPSLatitudeRef=South \
      -GPSLatitude=23.88 \
      -GPSLongitudeRef=West \
      -GPSLongitude=49.81 \
      pic.jpeg

With geotag-exif:

  geotag-exif pic.jpeg -23.88 -49.81

It won’t touch your EXIF tags other than the GPS tags and the FileLastModified timestamp.

The theme was “opposites”, and I’m interested in simulations, so my game idea was of an alien world with two complementary species. When pranjos die they create plants that are the source of food for bluos, and vice-versa. But these animals are not very adept at surviving, and their populations are threatened: oranjos are lazy and don’t explore much, so they starve to death without finding food; and bluos are timid and breed too seldom. As a space biologist, it’s your task to throw oranjos and bluos around to increase their numbers.

At least that was the idea.

[feed readers: continue to postmortem and screenshots…]

Structured procrastination rules

For some reason, working in this game made me very motivated to act on another idea I had for ages: a console roguelike using Japanese kanji and other Unicode characters. I made a few attempts at writing games in the past, but all came to naught. It’s hard to admit, but looking back, I think I simply lacked the experience and knowledge to finish one. I had no idea of how to write the essential basics, like how to model game objects and the world and where exactly to put their attributes and methods (should an object’s position be stored in the object itself, or in the world object? Who handles collision detection? Sight? Viewport scrolling reactions to object-induced events?). My math and geometry skills are rusty, and you need them like ALL THE TIME — even in 2D or console games — for such basic tasks as collision detection, proximity, line of sight, and so on. Trying to write games without paying enough attention to these topics confused me, and I’d interpret that feeling as “laziness” and stop coding.

So getting Maggaworld to do something was amazing, even if the “something” I did is only a bunch of oranjos walking over a single screen, getting older, breeding, and starving to death (which would be much more interesting had I draw more than three frames of animation). I didn’t make no plants, no bluos, no player, no real world with scrolling and uneven terrain. I actually only worked in Friday night and through the Saturday afternoon; then I got an irresistible urge to hack Makai・Jukai (the roguelike) — which I did in perfect flow until some ~800 lines of scheme mid-Monday, when I grudgingly teared myself away from home to, you know, work.

Maggaworld postmortem and links

The most costly decision was to use Chipmunk, that amazing 2D physics simulation library which IMHO indie developers don’t pay enough attention to. The demos are so cool, and I though I’d avoid calculating collision detection and gravity, but boy what a price to pay. 80% of the time I spent with Maggaworld wasn’t on the game itself, it was getting my head around the most basic principles of Chipmunk — it’s way too powerful for what I had in mind. In fact, I often had to learn concepts only to disable them, like setting the angular momentum of bodies to infinity to prevent rotation. In my first few attempts, I made the Shape definitions all wrong and the oranjos would just fall through the floor =D I ended up coding a falling squares demo to teach myself, and only then managed to get Maggaworld working. I documented the demo and hope it’ll be useful for other Gosu + Chipmunk users. If you’re also new to game coding and want to use Chipmunk in a weekend contest, I suggest writing some demos yourself before the start!

If Chipmunk’s complex (by necessity), Gosu is a pleasure. It’s a very ruby-like, convenient high-level library that makes sense; it’s how I wished SDL was when I first tried it. A shame neither Gosu nor Chipmunk are packaged in Debian yet (or in rubygems, for that matter).

Writing games is a lot of work! I seriously overestimated how much I could chew; the art alone took several hours (yes, those three 32×32 childlike sprites took all that time!). I stated that my goal with the contest was not winning but finishing something; at the end of the weekend I failed, but in another sense it was a big personal success — my ideas for how to code Makai・Jukai are coming to reality nicely, and I haven’t given up Maggaworld either (though I’ll probably add the missing elements at a leisure pace). I even forced myself to come out of the closet this time: there’s a rubyforge page for Maggaworld now, and a git repository for Makai・Jukai is available in my site at http://namakajiri.net/code/git/makai. I welcome input from anyone interested, and thanks all the people who kept up with my questions in freenode =D I’m totally joining the next contest!

While I’m at it: to respect the RFC, make sure your bots send messages with client.connection.notice and not client.connection.privmsg like many do. /notice is the same as /privmsg (aka /msg in most clients), only automated citizens ignore it. That prevents infinite loops, like the one I often trap myself with when talking to bitlbee’s root user:

   <leoboiko> root: yes
   <root> Did I ask you something?
   <leoboiko> root: yes
   <root> Did I ask you something?
   <leoboiko> root: yes
   <root> Did I ask you something?
   <leoboiko> root: yes
   <root> Did I ask you something?
   <leoboiko> root: yes
   <root> Did I ask you something?
   […]

• It takes ten years. For real.
• Books are necessary but not sufficient. You have to program, to read good programs, and to read bad programs.
• Computing ≠ computers ≠ programming.

But, FWIW, my ten books:

• GEB, for the same reason ricbit suggested: if you don’t like reading it, you won’t like computing (good one ricbit!)
• How to think like a computer scientist: Learning with Python. If needed, google for python tutorials to help.
• Data Structures and Algorithms in Python or any other algorithms book. Being python, this one has the advantage of succinctness.
• Structure and Interpretation of Computer Programs as the first computing book for adults. Ask help ;) Related reading: The little schemer & The reasoned schemer.
• Concrete mathematics: now we’re talking seriously. If I have to choose one math book for computer scientists, this gotta be the one.
• Computer architecture: a quantitative approach. It’s good to know a bit about computers.
• The C Programming Language, by the authors of C. This wonderfully-written little book is the antidote to Java and C++ behemots. Read with the architecture book to learn to deal with memory.
• Artificial Intelligence: A Modern Approach: Best intro to AI evar. AI techniques will do wonders to enlarge your horizons as a computing guy, and then you can enlarge them even more with…
• Probability and Computing: Randomized Algorithms and Probabilistic Analysis. Who said the best algorithms are deterministic?
• And The practice of programming, to polish your taste.

<jtg> Stupid python doesn’t have “resume” or whatever it’s called in ruby during a try…except clause.
<shc_> sigh
<leoboiko-onduty> implement it with continuations
<leoboiko-onduty> OH WAIT IT DOESN’T HAVE THEM EITHER
<leoboiko-onduty> </troll>

Irritado que o tracking dos correios não tem feed?

Amargo porque acha que nunca vão fazer um?

Cansado de apertar Ctrl+R no Firefox e dizer que sim, você quer reenviar os dados POST?

Seus pobremas acabaram!!

watch-encomenda.sh

New: suporte experimental a encomendas internacionals! (thx ademar)

Why do variable resistors come with nonlinear tapers? Well, as it turns out, human physiology has a weird way of perceiving changes in signal intensity, such as sound and light intensities. For example, you may think that if you doubled the intensity of sound or light, you would perceive a doubling in sound and light. Unfortunately—at least in terms of intuition (not in terms of safety control for our brains)—humans do not work this way. In fact, our perceptions of sight and sound work as follows: Perceived loudness/brightness is proportional to log₁₀ (actual intensity measured with a nonhuman instrument). Thus, if you are building an amplifier for a set of speakers or building a home light-dimming circuit, it would be wise to use a variable resistor with a nonlinear taper.

(From: Practical Electronics for Inventors, Paul Schers)

Wait.

> PING smeagol.vexcorp.com (200.160.255.84)
> smeagol.vexcorp.com
> smeagol

Oooh!

for i in aragorn gandalf frodo sam sauron melkor morgoth eru saruman balrog nazgul gollum; do
  host $i.vexcorp.com
  done|grep -v 'not found'
    gandalf.vexcorp.com has address 200.187.151.90
    sauron.vexcorp.com has address 200.160.255.100
    morgoth.vexcorp.com has address 200.160.255.101
    balrog.vexcorp.com has address 200.160.255.86

Turns out the log files are just a convention spelled out in /etc/syslog.conf (read syslog(3) and syslog.conf(5) if you don’t know syslog). These are the relevant lines in Debian defaults:

*.*;auth,authpriv.none      -/var/log/syslog

*.=info;*.=notice;*.=warn;\
    auth,authpriv.none;\
    cron,daemon.none;\
    mail,news.none      -/var/log/messages

The first line means: send all classes of messages (“facilities”) to /var/log/syslog, except the auth and authpriv facilities — these are sent to /var/log/auth.log instead (“auth” is just the deprecated name of “authpriv”).

The second line means: send all messages exactly at the levels of “info”, “notice”, and “warn” to /var/log/messages, except those from the listed facilities.

So /var/log/messages ⊂ /var/log/syslog; and, further, messages only contains generic non-critical messages. I have no idea why people use that, but there you go. If you want a complete log, you should look at /var/log/syslog and /var/log/auth.log.