Because I was using later versions of Mercurial and Python than those used in the guide (I was using Mercurial 2.1 and Python 2.6; there’s a note in the guide stating it used Mercurial 1.4.3 with Python 2.5.4, and should work with Mercurial 1.5 and Python 2.6.4), there were a couple of problems I had to workaround - if you’re using similar versions of Mercurial and Python to me then hopefully this’ll help you out, although YMMV:

1. Since Mercurial 1.6+, hgwebdir.cgi has been merged with hgweb.cgi. Therefore wherever the article mentioned hgwebdir.cgi, I swapped it for hgweb.cgi.

2. After extracting library.zip to C:\inetpub\wwwroot\hg, and copying the templates folder to C:\inetpub\wwwroot\hg, I tried to access http://name_of_server/hg/hgweb.cgi but Python threw an ImportError and complained it couldn’t find the mercurial module. To get round this I:

   • Reverted the previous copy and extract steps i.e. removed the templates folder, and the extracted library.zip, from C:\inetpub\wwwroot\hg
   • Downloaded mercurial-2.1.1.win-amd64-py2.6.exe from https://bitbucket.org/tortoisehg/thg-winbuild/downloads/ and ran the installer (NB you should download the .exe most appropriate for your server). This found the version of Python installed on my system and copied the relevant libraries / modules to Python’s site-packages folder. I was then able to browse to http://name_of_server/hg/hgweb.cgi without any problems.

If you’re new to web development, and new to version control, you should go and buy a copy.

I came across this problem the other week and wrote this little function in JavaScript to parse the JSON into a Date object:

Introduction to Node.js

New to this year’s event was the Afternoon with…Startups over at the Live Theatre, held the afternoon before the main conference. The venue was good, all the talks were really interesting and insightful - particularly those from Kevin Mann, Roan Lavery and Bobby Paterson - and I came away from it with plenty of new ideas and inspiration.

The next day was the main two track conference. The first talk of the day for me was Jake Archibald with @font-face - Good vs Legal. This was easily one of the highlights of the day. It was a brilliant talk with a brilliant slide deck brilliantly delivered (I would highly recommend watch the video of the talk when they’re published shortly). He even used a Wii remote to change move onto the next slide :)

I stayed on the Build It track for the rest of the morning. Bradley Wright gave an insightful look into Programming the Real-Time Web. He covered the various ways that real time info can be sent and displayed in the browser, from the forever iframe approach to newer methods such as those provided by WebSockets and Socket.IO, and also how you can keep your real time web app accessible by using WAI-ARIA attributes.

Rounding out the morning session was Rich Thornett on Using NoSQL with Yo’SQL. It was an interesting look at how Dribbble makes use of MongoDB, where they’ve favoured it over relational databases and on how it can be used to eliminate N + 1 queries to help boost your site’s performance.

First up after lunch was Corey Donohoe, with another one of the day’s highlights for me. It was a fascinating talk on all the metrics they use at GitHub to help them deploy changes to their site really, really quickly. Front-end behaviour, back-end behaviour, performance and failure rates are all monitored using a variety of tools. Plus there was plenty on the awesomeness that is Hubot!

Next I hopped over to the Design It track for Jeremy Keith and his One Web talk. It was probably the most inspiring talks of the day. It focused on how the “mobile web” is a misnomer and shouldn’t be treated as a separate entity, but instead by adopting a content first approach and focusing on universal access to the web from any device, you can end up with a much better, scalable and richer user experience.

After an exceptionally well timed Red Bull break, I headed back to Build It to catch Blaine Cook talk about Building the Social Kaleidoscope. It featured a really interesting look at Webfinger, what it is, how it works and how it can help to solve the problems identifying users on the Web when they’re details are stored in multiple places and walled gardens).

The final talk of the day was the keynote from none other than Jeffrey Zeldman. Entitled What Every Web Designer Should Know, it was a very entertaining talk on the importance of designing from the content out, defining the problem before you try and solve it and, in a similar vein to Jeremy Keith, on how designing for mobile devices first can help you focus on what the user wants to do.

Overall DIBI was a fantastic conference, all the speakers were great and I came away feeling like I learnt a lot. Huge plaudits must go to Gavin Elliot and the team for all their hard work. Already I’m really looking forward to DIBI 2012. [DIBI]: Design It Build It [WAI-ARIA]: Web Accessibility Initiative - Accessible Rich Internet Applications

By default if uses base58 encoding, but it does let you pass it your own base alphabet allowing you to do base16, base32 or base-whatever-you-like encoding.

Please have a play with it and let me know what you think :)

HTML5 Form Validation

• Don’t rely on other people to remove your totally fake “distractions.”
• Don’t rely on other people to pat your beret, re-tie your cravat, and make you a nice cocoa whenever that mean man on the internet points out that your “distractions” are totally fake. (Which they are)
• Don’t rely on other people to tell you when or whether you have enough information.
• Don’t rely on other people to define your job.
• Don’t rely on other people to “design your lifestyle.”
• Don’t rely on other people to decide when your opinions are acceptable.
• Don’t rely on other people to tell you when you’re allowed to be awesome.
• Don’t rely on other people to make you care.
• Don’t even rely on other people to tell you what you should or shouldn’t rely on.

From http://www.43folders.com/2010/10/05/distraction

if you ask a “what if?” question and leave it unanswered, you should be ashamed

From http://ajaxian.com/archives/web-ninja-interview-marcin-wichary-creator-of-google-pacman-logo-html5-slide-deck-and-more

Basically, the problem was being caused by me using an external jar file in my project and unit tests. There were 2 steps I needed to follow to fix things:

1. Export the external jar file from the main project i.e. not the test project
2. Remove the reference to the external jar file from my test project

Et voila - my unit tests were up and running. Getting them all to pass was another matter though :)

Feel free to fork it or offer suggestions for improvements in the comments :)

Two such tips that I picked up this week are:

1. When using the throw statement you can use an object literal to give an exception a type and a message. For example:

throw {
  type: 'FooError',
  message: 'Invalid foo'
};

1. When using closures, avoid creating an anonymous function inside a loop as it can be computationally wasteful and potentially lead to confusing code. For example (this is taken from the book):

var add_the_handlers = function (nodes) { 
    var helper = function (i) {
        return function (e) { 
            alert(i);
        };
    }; 

    var i; 
    for (i = 0; i < nodes.length; i += 1) {
        nodes[i].onclick = helper(i);
    }
};  

Anyway, here they are: Visual Studio settings files

mysql -u foo -p dbname < script.sql

When running this the other day though I was presented with the following error:

ERROR 1153: Got a packet bigger than ‘max_allowed_packet’ bytes

Fortunately help was at hand via this forum post. All I needed to do was run some SQL statements to set some global variables in MySQL like so:

set global max_allowed_packet = 1000*1024*1024; 
set global net_buffer_length = 1000000;

To check whether this worked was merely a matter of running:

-- should return 1048576000
select @@max_allowed_packet;

Of course, you could always set the max_allowed_packet in your my.cnf file instead.

Update

After upgrading my MacBook Pro to OS X Snow Leopard, error 1153 was replaced with:

ERROR 2006: MySQL server has gone away

Fortunately the above still fixed things. Of course, if I’d listened to Phil Sherry in the first place I might not have had to re-install MySQL after upgrading OS X (although installing MySQL via the binary package installer did seem to install in /usr/local/bin leaving me puzzled as to why it disappeared after the Snow Leopard upgrade. But that’s a story for another day…). *[SQL]: Structured Query Language

This no longer works with the new GMail UI

I quite like the toolbar at the top of GMail. You know, the one with links to Google Calendar, Google Reader, Google Docs, et al. But I’m not that keen on having to scroll right back to the top of the page each time I need access to it. So, I created a simple Greasemonkey script for Google Chrome to apply fixed positioning to the toolbar. That way it always stays at the top of your screen :)

(Firefox users: I did create a user stylesheet that achieves the same effect recently. However, that relies on the @-moz-document domain(mail.google.com) to restrict the CSS to GMail and Google seems to send different HTML to Firefox than it sends to Chrome).

You can fork a copy of the script from GitHub: https://github.com/ianoxley/gmailtoolbar

Or alternatively download it from here: gmailtoolbar.user.js

It’s a Bash script that lets you export a list of Google Docs to a specified format. For example, to export a particular folder to PDF you’d type something like this:

googlecl-export -f pdf -d yourfolderofdocs

You need to have GoogleCL installed for the Bash script to work (GoogleCL can be downloaded from http://code.google.com/p/googlecl/) but otherwise you should just need to make the script executable e.g. chmod u+x googlecl-export and possibly stick it somewhere on your $PATH

But, rather than manually adding and removing this class to the page.tpl.php, I created a bookmarklet to toggle the show-grid class on and off. The code looks like this:

(function() {
  if (document.body.className) {
    var b = document.body;
    if (b.className.indexOf(' show-grid') !== -1) {
      b.className = b.className.replace(' show-grid', '');
    } else {
      b.className += ' show-grid';
    }
  }
})();

It does make the assumption that your <body> tag will always have a class attribute, although it should be fairly straight forwards to modify the code to allow for cases where the <body> tag does not have a class attribute.

To save it as a bookmarklet, drag this link to your bookmarks bar: Show Grid%20{if%20(document.body.className)%20{var%20b%20=%20document.body;if%20(b.className.indexOf(%27%20show-grid%27)%20!==%20-1)%20{b.className%20=%20b.className.replace(%27%20show-grid%27,%20%27%27);}%20else%20{b.className%20+=%20%27%20show-grid%27;}}})();)

To do this in Drupal 6 turned out to be a lot easier than I thought. I used the code from this blog post by Nicholas Coates but, rather than add it to a views template, I added it to a hook_views_query_alter function I was already using. This meant I could easily use the $view object passed as a parameter to this function:

function hook_views_query_alter(&$view, &$query) {
  ...
  $itemCount = $_GET['itemcount']; // this is just as an example
  if ($itemCount == 0) {
    $view->set_items_per_page(10);
  } else {
    $view->set_items_per_page($itemCount);
  }
    ...
}

As Nicholas Coates says in his aforementioned blog post: this is a quick and painless way to implement an items per page filter.

  "Can't create/write to file '/path/to/folder/filename' (Errcode: 2)"

I was having this problem the other day and, after checking that file permissions weren’t the cause of the problem, I came across the following post on the MySQL forums:

  "The problem was not in MySQL but in AppArmor on Ubuntu. I had to add the directories I wanted to write into to my /etc/apparmor.d/usr.sbin.mysqld and restart apparmor.d."

So, one sudo vim /etc/apparmor.d/usr.sbin.mysqld later and I’d added the path/to/folder I needed to be able to write to (not forgetting to add the trailing comma ‘,’ to the end of the line, which is always required, even for the last line).

All that was left to do was to restart AppArmor with:

$ sudo /etc/init.d/apparmor restart

After that, SELECT…INTO OUTFILE worked like a charm :)