multipled = [3 * s for s in series]

One list comprehension later, the task is complete.

Next, we find out that we need to count how many multiplications are performed. len() is forbidden, because we are dealing with iterators (and inevitably the itertools module). One solution is to create a CallCounting class that wraps callables.

class CallCounting(object):

    def __init__(self, func):
        self.count = 0
        self.func = func

    def __call__(self, *args, **kwargs):
        self.count += 1
        return self.func(*args, **kwargs)

timesthree = CallCounting(lambda x: 3 * x)
multipled = [timesthree(n) for n in series]

print "%d calls" % timesthree.count

This is efficient as well as readable. The main action is to multiply items. We keep a call count just for bookkeeping. Compare with

calls = 0
multipled = []
for n in series:
    calls += 1
    multipled.append(3 * n)

print "%d calls" % calls

that is both ugly and inefficient.

if (!device || !this.launched) {
    abort();
}
else {
    run();
}

with

if (device && this.launched) {
    run();
}
else {
    abort();
}

Chances are, you are wired to parse the second version faster. Make it a habit to see if your code’s readability can be improved by avoiding negations.

Nesting can also improve readability.

if (a) {
    doFirstThing();
}
if (a && b) {
    doSecondThing();
}
if (!a && b) {
    doThirdThing();
}

versus

if (a) {
    doFirstThing();
    if (b) {
        doSecondThing();
    }
}
else {
    if (b) {
        doThirdThing();
    }
}

The best logic of all is no logic. Compare

if (person.isChild) {
    run();
}
else if (person.isBaby) {
    crawl();
}
else if (person.isSenior) {
    sit();
}
else {
    walk();
}

with a data-driven approach

var groupAction = {
    children: run,
    babies: crawl,
    seniors: sit
}

var action = groupAction[person.group] || walk;
action();

See this page for details of 1927 in Shanghai. Basically at midnight at the end of 1927, the clocks went back 5 minutes and 52 seconds. So “1927-12-31 23:54:08” actually happened twice, and it looks like Java is parsing it as the later possibly instant for that local date/time - hence the difference.

Just another episode in the often weird and wonderful world of time zones.

Besides illustrating how difficult dealing with time is, this is a textbook example of why estimations really are guesstimations. I am certain that Freewind spent a good deal of time trying to solve this problem before submitting it to Stack Overflow. In any project of considerable size, you run into these road bumps. Even though you know how of a workaround, it is worthwhile to find the root cause and solve problems in a clean manner instead of polluting systems with ad-hoc fixes. Given enough time, any quick-fix will come back to haunt the maintenance programmer.

It works pretty much out of the box in Django, but a little configuration is needed.

Programs must be written for people to read, and only incidentally for machines to execute.

window['callback'+idx] = callback;

and then ?jsonp=callback0, ?jsonp=callback1, …

The advantages of doing things in software on a single device are so great that everything that can get turned into software will. So for the next couple years, a good recipe for startups will be to look around you for things that people haven’t realized yet can be made unnecessary by a tablet app.

Other companies may indeed force 45-50 hour work weeks plus weekends, cram them in elbow-to-elbow and give them health plans with a $3,500/year deductible, but it doesn’t mean you can sit pretty with your lead developer grinning and taking it. There’s four things a developer, like any other human being, wants from a job: competitive salary, comfortable environment, challenging work, and a future. If you keep filling vacancies at the top from outside the company, then the bright kids at the bottom will seek work elsewhere even if it pays the same but gives them a sense of having a career path. If you make them work in a closet with 4 other programmers and inadequate ventilation, they’ll bolt for the first place that gives them an office with a window. If you hire a whiz-kid and make him edit Microsoft Access forms in Visual Basic, you won’t see him after a few months. If they find another job that pays more but has longer hours and commute, they will take it because it’ll make them feel like they’re getting paid what they’re worth.

Of course, you could just hire H1B workers exclusively. That’ll last until they get citizenship, or the other country’s economy rises above ours.

The simplest thing that we can do right now is to rethink how we document libraries. Automatic documentation tools like Javadoc and Rdoc are good for producing thick stacks of paper, but not so hot at actually telling library users the things they need to know. They are the source of most of the world’s “newRecordAddedHook is called when a new record is added” documentation. They can be useful, but they are no substitute for actually writing about the library: single-page summaries that answer the three key questions: what the library does, why you should use it, and how to do so. For bonus points, the one-page summary should avoid using the word “enterprise” (unless it’s a library of Star Trek ships) and “innovative.”

The only difference between a bug and a feature is the documentation.

Source unknown.

The romantic image of an über-programmer is someone who fires up Emacs, types like a machine gun, and delivers a flawless final product from scratch. A more accurate image would be someone who stares quietly into space for a few minutes and then says ‘Hmm. I think I’ve seen something like this before.’

The ultimate goal of computer science is to help produce better systems. Would you trust someone who had not seen a patient for years to teach surgery? What would you think of a piano teacher who never touched the keyboard? A CS education must bring a student beyond the necessary book learning to a mastery of its application in complete systems and an appreciation of aesthetics in code.

You wanted a banana but what you got was a gorilla holding the banana and the entire jungle.

Professors often provide a page with links to lecture slides in PDF format. The main benefit of joining them to a single document, is that one can use a reader’s search function to find topics and keywords faster than when searching in multiple documents.

The Script

Change the URL variable to the page with links to lecture slides. Edit the grep and pdfjoin lines to match filenames of provided lecture slides.

#! /bin/sh

TARGET_FILE=all-lectures.pdf
URL="http://www.ida.liu.se/~TDDC88/theory/lectures.shtml"

for prog in wget lynx pdfjoin; do
    which $prog 1>/dev/null
    if [ $? -ne 0 ]; then
        echo $prog needed but not found.
        exit 1
    fi
done

PDF_URLS=$(
    lynx -listonly -dump -hiddenlinks=merge $URL \
    | tail -n+4 \
    | awk '{print $2}' \
    | grep 'lecture-.*-pps6.pdf'
    )

TEMPDIR=$(mktemp -d)
cd $TEMPDIR
echo Fetching PDFs...
wget $PDF_URLS
echo Joining documents...
pdfjoin $(seq -f 'lecture-%g-*-pps6.pdf' 1 $(echo $PDF_URLS | wc -w))
cd -
mv $TEMPDIR/*-joined.pdf $TARGET_FILE
rm -r $TEMPDIR
echo PDF $TARGET_FILE was generated.

Future Improvements

• Get rid of the lynx dependency
• Download all found documents simultaneously (in background jobs, perhaps)

Some features of Real VMX:

• Priority-based multitasking and round robin scheduling
• Partition-based memory management
• Binary and counting semaphores with priority-inheritance
• Message queues for inter-process communication
• Virtual memory

Real VMX is licensed under the LGPL.

Get the script here! (only tested in Linux)

countdown.c:

#include <stdio.h>
#include <unistd.h>

int main(int argc, char **argv)
{
    int seconds = 3; 

    for (int i = seconds; i != 0; --i) {
        printf("%i... ", i);
        fflush(stdout);
        sleep(1);
    }
    puts("Go!");

    return 0;
}

The goal is to put the actual countdown code in a macro. It fits perfectly in a function, but a macro is used for illustration purposes. A first attempt at a solution is presented.

solution1.c:

#include <stdio.h>
#include <unistd.h>

#define COUNTDOWN(seconds) \
    for (int i = (seconds); i != 0; --i) { \
        printf("%i... ", i); \
        fflush(stdout); \
        sleep(1); \
    } \
    puts("Go!")

int main(int argc, char **argv)
{
    COUNTDOWN(3);

    return 0;
}

This works fine in the example, but consider what happens when an if clause is added.

Part of solution1-if.c:

    bool start_countdown = false;

    if (start_countdown)
        COUNTDOWN(3);

The output when this program is run is “Go!.” To know what happens, look at the output of gcc -std=c99 -O0 -E solution1-if.c | indent -linux:

int main(int argc, char **argv)
{
    _Bool start_countdown = 0;

    if (start_countdown)
        for (int i = (3); i != 0; --i) {
            printf("%i... ", i);
            fflush(stdout);
            sleep(1);
        }
    puts("Go!");

    return 0;
}

Oops! A second attempt is to wrap the macro in braces.

Part of solution2.c:

#define COUNTDOWN(seconds) \
    { \
        for (int i = (seconds); i != 0; --i) { \
            printf("%i... ", i); \
            fflush(stdout); \
            sleep(1); \
        } \
        puts("Go!"); \
    }

Perfect! This code runs as expected.

Now consider what happens when an else clause is added.

Part of solution2-else.c:

    if (start_countdown)
        COUNTDOWN(3);
    else
        puts("Not ready to start countdown.");

This doesn’t even compile!

$ gcc -std=c99 solution2-else.c
solution2-else.c: In function ‘main’:
solution2-else:21: error: ‘else’ without a previous ‘if’

Again, have a look at the code after macro expansion:

int main(int argc, char **argv)
{
    _Bool start_countdown = 0;

    if (start_countdown) {
        for (int i = (3); i != 0; --i) {
            printf("%i... ", i);
            fflush(stdout);
            sleep(1);
        } puts("Go!");
    };
    else
    puts("Not ready to start countdown.");

    return 0;
}

Notice the badly placed semi-colon. A final solution is presented, which will behave as expected in all situations.

final.c:

#include <stdbool.h>
#include <stdio.h>
#include <unistd.h>

#define COUNTDOWN(seconds) \
    do { \
        for (int i = (seconds); i != 0; --i) { \
            printf("%i... ", i); \
            fflush(stdout); \
            sleep(1); \
        } \
        puts("Go!"); \
    } while (0)

int main(int argc, char **argv)
{
    bool start_countdown = true;

    if (start_countdown)
        COUNTDOWN(3);
    else
        puts("Not ready to start countdown.");

    return 0;
}

After macro expansion:

int main(int argc, char **argv)
{
    _Bool start_countdown = 1;

    if (start_countdown)
        do {
            for (int i = (3); i != 0; --i) {
                printf("%i... ", i);
                fflush(stdout);
                sleep(1);
            } puts("Go!");
        } while (0);
    else
        puts("Not ready to start countdown.");

    return 0;
}

One concern that needs to be addressed is how the seconds argument could be expanded. Furthermore, this technique isn’t the preferable solution most of the time, but if you run into it, you’ll know what it’s usually for. Another use is to prevent a macro to be used as an expression,

#define INCR_EXPR(x) ++(x)
#define INCR_STMT(x) do { ++(x); } while (0)

int main(int argc, char **argv)
{
    int a = 0;
    int b = 0;

    a = INCR_EXPR(b);
    INCR_STMT(b);
    return 0;
}

had it been,

    a = INCR_STMT(b);

gcc would fail with an error “expected expression before ‘do’:”

a = ++(b);
a =
do {
    ++(b);
} while (0);
return 0;

A Simplified Theme

<html>
<head>
  <title>My Tumblr</title>
  <script type="text/javascript" src="http://example.org/jquery.js"></script>
  <script type="text/javascript" src="http://example.org/tumblr-dyn.js"></script>
</head>
<body>
    {block:Posts}…{/block:Posts}
    <p id="quote-of-the-day" style="display:none"></p>
</body>
</html>

We shall add a quote to #quote-of-the-day for permalink views.

tumblr-dyn.js

var isSinglePostView = function() {
    var loc = document.baseURI;
    var re = new RegExp(".*mytumblr.example.org/post/.*");
    return loc.match(re) != null;
}

$(window).ready(function() {
    if (isSinglePostView()) {
        var qotd = $("#quote-of-the-day");
        qotd.html(
            '"You should\'ve seen the look on her face. It was the same look ' +
            'my father gave me when I told him I wanted to be a ventriloquist."' +
            ' – George Costanza');

        qotd.css('display', 'block');
    }
});

You can use this approach to add custom content depending on what page the user is viewing.

Consider the following network,

where Alice and Bob both need to work with two servers, on the same external IP address, and a restrictive firewall (or sysadmin) that refuses to open non-standard ports for the internal servers’ SSH daemons. However, the SSH daemon on a third internal host, the SSH gateway, is reachable by Alice and Bob from the outside. The SSH gateway can connect to the web and mail servers’ SSH daemons by their internal IP addresses.

Now, Alice and Bob could be satisfied with connecting to the SSH gateway, and then to manually open a second connection to the web or the mail server. This is tiring in the long run, however, and the rest of this article describes how to automate this process.

First, two new user accounts are created on the gateway: web-server, and mail-server.

$ sudo su -
# useradd web-server
# useradd mail-server

We run the ssh-keygen utility for both new users, to create public/private SSH key pairs in the ~/.ssh directory. These key pairs will be used to let Bob and Alice connect from the gateway to the internal servers, without having to enter their passwords.

$ sudo su web-server -
$ ssh-keygen
Generating public/private rsa key pair.
Enter file in which to save the key (/home/web-server/.ssh/id_rsa):
Created directory '/home/web-server/.ssh'.
Enter passphrase (empty for no passphrase):
Enter same passphrase again:
Your identification has been saved in /home/web-server/.ssh/id_rsa.
Your public key has been saved in /home/web-server/.ssh/id_rsa.pub.
The key fingerprint is:
cf:9e:18:a3:dc:6c:48:1a:b4:7d:5c:d5:84:da:d0:f1 web-server@gateway
The key's randomart image is:
+--[ RSA 2048]----+
|           ..=.  |
|          . +..  |
|           =  E  |
|    .     o .    |
|   . o .S.       |
|    o o oo       |
|     + oo o      |
|    ...+.= .     |
|      o.+ o      |
+-----------------+
$ exit

(Repeat this twice, the second time for mail-server.)

It is assumed that Alice and Bob already have their own accounts on the servers. The next step is to configure these accounts to authenticate connections from the gateway by keys.

1. Copy public keys on the gateway to the servers

$ sudo su web-server -
$ scp ~/.ssh/id_rsa.pub alice@web-server:web_key
The authenticity of host 'web-server (192.168.0.10)' can't be established.
RSA key fingerprint is 72:1d:15:7a:05:21:f3:fe:e2:7a:11:88:ab:45:18:bd.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added 'web-server' (RSA) to the list of known hosts.
alice@web-server's password:
id_rsa.pub                                             100%  393     0.4KB/s   00:00
$ exit
$ sudo su mail-server - # now the same thing on the mail server
$ scp ~/.ssh/id_rsa.pub alice@mail-server:mail_key
The authenticity of host 'mail-server (192.168.0.20)' can't be established.
RSA key fingerprint is 72:1d:15:7a:05:21:f3:fe:e2:7a:11:88:ab:45:18:bd.
Are you sure you want to continue connecting (yes/no)? yes
Warning: Permanently added 'mail-server' (RSA) to the list of known hosts.
alice@mail-server's password:
id_rsa.pub                                             100%  393     0.6KB/s   00:00

2. Add public keys to ~/.ssh/authorized_keys on the servers

$ ssh alice@web-server
alice@web-server's password:
$ mkdir -p .ssh
$ cat web_key >> .ssh/authorized_keys
$ chmod go-rwx .ssh/authorized_keys
$ exit
$ ssh alice@mail-server # same thing at the mail server
alice@mail-server's password:
$ mkdir -p .ssh
$ cat mail_key >> .ssh/authorized_keys
$ chmod go-rwx .ssh/authorized_keys
$ exit

(Repeat step 2 for Bob.)

The result of step 1 and 2 is that the users web-server and mail-server on the gateway can open connections to their internal targets without having to enter a password. Only users who are trusted to use the servers as both Alice and Bob shall be trusted to log in directly as web-server or mail-server at the gateway.

The final step is to let Alice and Bob authenticate as web-server and mail-server on the gateway with their external hosts’ public keys, and to make the gateway automatically open a second connection to the appropriate internal server.

1. Get Alice’s and Bob’s public keys (found in ~/.ssh/id_rsa.pub or possibly ~/.ssh/id_dsa.pub on their machines). If the keys haven’t been generated, ask them to run ssh-keygen as described earlier.
2. Add both Alice’s and Bob’s keys to /home/web-server/.ssh/authorized_keys and /home/mail-server/.ssh/authorized_keys on the gateway. This step is described earlier.

3. Prepend “command” instructions to lines in the authorized_keys files, on the gateway. Next is an example on how /home/web-server/.ssh/authorized_keys could look.

   $ cat ~/.ssh/authorized_keys
   command="ssh -t bob@web-server" ssh-dss AAAAB3N…YJylbjCVHQ== bob@bob-external-host
   command="ssh -t alice@alice-server" ssh-rsa AAAAB3N…a9Sydw== alice@alice-external-host
   

To try it out, ask Alice or Bob to authenticate as web-server or mail-server on the gateway. Here’s how it shall look for Alice:

$ ssh web-server@gateway-external-host
Last login: Fri Oct  9 11:56:59 2009 from alice-external-host
$ whoami
alice 
$ exit
Connection to web-server closed.
Connection to gateway-external-host closed.
$ ssh mail-server@gateway-external-host
Last login: Fri Oct  4 18:56:59 2009 from alice-external-host
$ whoami
alice 
$ exit
Connection to mail-server closed.
Connection to gateway-external-host closed.