I’ve been playing with the stackoverflow beta since Saturday. Stackoverflow is a project by Jeff Atwood (of CodingHorror fame), with moral support by Joel Spolsky (JoelOnSoftware). Together they’ve been doing a podcast for almost four months. Stackoverflow launched a private beta last Friday.

frontpage

Stackoverflow has a fairly simple concept. Users can ask questions about programming related topics, other users answer them. The questioner may mark one or more answers “accepted” - these answers move to the top of the list and are marked in green.

Anyone may up- or downvote questions and answers - this affects the order of the answers. There are separate lists of the most recent questions, the most popular questions, and questions with the least answers. Questions also have tags (mostly programming languages and frameworks), and users can browse questions per tag.

Currently, with only a couple of hundred users, the site works reasonably well. If I ask a question that isn’t too technical (unless it’s .NET - the userbase is heavily skewed towards the .NET platform) I usually get several decent to good answers within the hour - afterwards, the responses taper off dramatically. The edit box uses markdown, with good syntax highlighting for most programming languages. There’s a live javascript-based preview, which works very well. It’s not yet possible to subscribe to interesting questions. Your own questions and answers appear on your profile page, but even there it’s hard to see which questions have received new answers.

Tag page

Stackoverflow tries to be a forum and a wiki at the same time, and uses a single abstraction (the question + answers) for two different things (a forum thread and a wiki article).

That’s ambitious, and I’m not sure if it’s going to work. There are already complaints that by sorting the answers by score, it’s impossible to follow the flow of the discussion. Having multiple sorting options may not be the answer. Also, because the software presents replies as answers to the question and doesn’t have threading, it somewhat inhibits a free-flowing discussion, with counter-questions, requests for clarifications, arguments, etc.

Question Page

Also, Stackoverflow has a couple of wiki-like features which haven’t quite kicked in yet. For one, when users hit a certain karma threshold, they may edit anyone’s question or answer. Versioning, reverting and diffs are also planned. For now, only a handful of users have reached the threshold, but I wonder what effect this will have on the social interactions on the site.

It’s a bit premature to criticize a site that’s been live for five days. The developers have been extremely active, and have already implemented many suggestions from the community (using the excellent and highly recommended UserVoice customer feedback site).

But I believe that the original intention was to be a more wiki-like resource, with time-tested answers to popular questions, and it isn’t like that at all now - it feels much more like a forum, with a heavy dose of voting and karma thrown in. It will be interesting to see whether the developers are agile enough to steer it toward the original goal.

I wish them the best of luck.

I have electronic scales, which allow you to enter your age and sex. When I step on my scales as a 33 year old man, I weigh 72.5 kg. Anonymously, however, I only weigh 70.7 kg.

My curiosity was piqued, of course, and I did two more experiments by placing a towel over the scales (the scales measure my body fat with a small electric current - the towel blocks the current). No difference as anonymous person, but as a 33 year old man without body fat measuring I weigh 70.7 kg.

So based on the measurement of my body fat and my age and gender, my scales add another 1.8 kg to my weight. I have no idea why.

The images, css files and javascript files in a web application, the “static resources”, are a big part of each request, both in number and in bandwidth. That’s why they they should be cached by the browser for as long as possible. By setting a HTTP-EXPIRES header in the far future (at least a year from now) the web server tells the web browser that it’s ok to cache these resources, saving us a lot of bandwidth and improving the page load time for the user.

But sometimes we want to make a change in one of our resources, and want the browser to update the resource at once, long before the cached version expires. To do this we must create a new URL for the resource. Yahoo solves this in their YUI project by giving their css resources a version number:
<link rel="stylesheet" type="text/css" href="http://yui.yahooapis.com/2.5.2/build/reset/reset-min.css"/>
This is a fine solution, but it involves some overhead - if you just want to use the latest version, you have to update the version number in all places you include the resource.

I’ve invented another trick which works fine as long as we, unlike Yahoo, don’t care about supporting older versions of resources. To generate the URL of a resource, we call a function, e.g. versioned_css("main"). In this function we append the git hash (git-hash-object filename) to the filename, before the extension, caching the result (because we don’t want to calculate the hash more than once). The first ten digits of the git hash will be sufficiently long to be unique (a git hash consists of 40 hexadecimal digits). Also, there is no need to use git - there are other ways to create a unique hash from a file.

Whenever the file changes, its hash will also change, creating a new and uncached URL, which is exactly the effect we’re looking for.

However, we still need to map the URL-with-hash to our resource, a file named something like “main.css”. We could add the hash to the filename, by renaming the resource to “main-c0a9a0a5ec.css” during the deployment of the application, but it is easier to have Apache remove the hash using url-rewriting:
RewriteRule ^/stylesheets/(.+)-[0-9a-f]{10}\.css$ \
http://127.0.0.1:3010/stylesheets/$1.css [P,QSA,L]
The difference with renaming the resource is that with rewriting a request for “main.css” with any hash at all will return the latest version of the file, while with renaming an incorrect hash would lead to a 404 error.

The strength of this trick is that it lets you use far-future expire dates, is easy to implement and allows you to forget about it afterwards, particularly if you extend the built-in methods in Rails to generate urls for resources.

Is your program as fast as it needs to be? As fast as it can be? Does it use as little memory as possible? Is it free of memory leaks? Is the response time that the users experience as low as possible? Is their a cap on the maximum response time? A low cap? Does your program consume as little bandwidth as possible? Does it consume no more energy than necessary?

Can your program report on its own functioning? Can it detect its own errors? Report them? Can you change your program while it’s running? Can your program update itself? Securely? In the background? Without restarting?

Is your program easy to understand for new programmers? Easy to change? Can parts of it be changed or configured by non-programmers? While running? Can you easily find programmers that are proficient with the languages and technologies you have chosen? If new programmers introduce errors, can these errors be easily detected? How does your program handle faulty user input? Deliberately invalid user input? Do you make a separation between parts that will probably remain unchanged and parts that will probably see a lot of change? Can your program be built in one step? Can you create and distribute patches for older versions?

Can parts of your program be used in other programs? Replaced by other modules? Can a single module be assigned to another team for improvement?

Can your program be scripted on the command line? Does it expose an API? Are there bindings or wrappers for Java and PHP? For Ruby and Python? For OCaml and Smalltalk?

Does the performance of your program increase linearly with faster hardware? With more hardware? Can you sacrifice memory usage for speed and vice versa with one setting?

Can your program survive a computer crash? Without data loss? Without downtime? Does it have a single point of failure? If it crashes, will you be notified automatically? At once?

Does your program run on IE, Firefox, Opera and Safari? All versions of IE? On windows, mac and linux? On windows 95 and Vista? Without administrative privileges? On all common linux distributions? On 64-bit hardware? On a PowerPC? On an iPhone? With a Turkish user interface? A Japanese interface? Will it work for the colorblind? The visually-impaired? The illiterate?

Does your program follow code conventions? Does it have conventions in naming variables and functions? Have you documented them? Does it have a glossary of terms? Can your program be read from beginning to end as a literate program? Understood? Understood a year from now by you? By other people? Does the documentation answer all the relevant questions in this essay?

Is your program correct? Is what it does worth doing?

Inglua now supports the French in their neverending quest to learn foreign languages. Speakers of French may now choose to learn English, German, Italian, Spanish or Dutch - the five languages of their neighbors.

This brings the total of language combinations that Inglua supports on 20. Next up is Spanish.

Inglua used to be very picky. Whenever you’d enter a very long sentence with 2 nouns, 2 adjectives and a tough verb, it would grade it as a fail even if you’d just made a single typo.

Now, it will grade the sentence parts separately. That means that if you do not know all the words in a sentence, it will still pay off to type whatever you do know, because you will get points for them!

Enjoy practicing!

I’ve been working very hard on this for the last three months, but I’m done. Italian is now one of the six supported languages.

For the time being, no new languages will be added. With the current selection, Inglua covers the bigger part of Europe and the Americas. It is now time to prove that the concepts works before moving into new markets.

I’ve been tempted for other reasons though. All current languages are fairly alike. Adding a completely unrelated language (such as Japanese) to the mix would certainly expose some of the cracks in my current approach.

I’ve just fixed Bug 34. The most recent bug in my bug-tracking system is No. 2654, so I’ve been taking my time. It was a tricky bug, but having solved it at last feels very rewarding.

It was also an interesting bug, and had to do with some peculiarities of English grammar. Usually, you can easily turn a sentence that’s in the present into the past or the perfect tense. “I see”, “I saw”, “I’ve seen”. But in English, there are some exceptions, and they’ve got to do with modal verbs, like “must”, “can”, and “may”.

He says that I may swim.

*He said that I might swim.

*He says that I have might swim.

“May” at least has a past tense, but you can’t use it, because there’s a shift in meaning between “may” and “might”. And neither “may”, nor “can”, nor “must” has a perfect tense. So what do you do? You transform the sentence by using another construction.

He says that I have been allowed to swim.

And that’s exactly what Bug 34 was all about. It now works correctly, and the title of this post can now be translated into all other languages.

It took me a long time, but Inglua has reached another important milestone: The vocabulary of Italian words is now complete!

There is still some ambiguity, so some 100 words need examples, and the grammar is still less advanced than the grammars of the other languages. When that is done, I’ll make Italian available as an option in the menu. Nevertheless, this is a huge step!

Arrivederci, amici! A presto!

You’ve probably wondered what Inglua stands for. Well wonder no longer! The Secret Is Revealed.