Following the HyperDB talk, Jeremy Zilar talked a bit about the blogs at New York Times that he maintained. It is interesting to note that they are running all the blogs on a single install of WordPress 2.0.5.

Rashmi Sinha’s talk, Designing Massively Multiplayer Social Systems, focused on the general topic of social networking. I think she made a good point about the importance of choosing the right popularity metric for the right purpose. Slideshare can sort the slides according to Latest/Most Viewed/Featured/Most Commented/Most Favorited/Most Zinged/Most Downloaded/Contest, each represents a popularity metric for a certain purpose. It is quite useful to me as well, since we are designing a blogging site for the Chinese bloggers recently.

In the afternoon talks, I have to mention the talk by Liz Danzico of Happy Cog Studios, Usability Analysis of WP. She is helping the WordPress team to re-design the next generation administration pages based on the usability analysis done on many WordPress users. During the talk, she mentioned that the Dashboard interface would be redesigned to allow customization using AJAX drag and drop features and possibly turn off the developer RSS feed and add a new feed from somewhere else. The menus will also get re-organized so that it would be easier for new users to access to certain commonly used pages. Because of the history of WordPress, the administration panel in WordPress is quite outdated and really needs a new look and re-organization. Considering that these changes to the backend administration pages should be expected in WP 2.4 release (probably by the end of the year), the wait for a better admin panel will be over soon.

Matt Mullenweg gave the most anticipated talk of the day, State of the Word. I loved his style of presentation and his honesty of pointing out some of the issues that they have not resolved. The developer team of WordPress has now adopted a better release scheme and promises to do 4 major release every year. In the next version of WP 2.3, tagging will be included, along with some features of plugin update notification, based on the central plugin repository on WordPress.org. WP 2.4 will likely to include the admin panel enhancement and possibly a core file update system (think about a button to upgrade your WP install).

Anyway, the WordCamp event ended with the Developer Duke-out, where Matt and other core developers, went up stage and talked about various issues and mostly fun topics related to WordPress. For example, they talked about their most-hated features of WordPress(haha), and the comment review and the image upload feature were on the top list. They also talked about their favorite text editor: most of them actually use nano under Linux. I have to say most of the developers are very geeky looking, and a bit shy on the stage, except Matt and Andy.

Now, let me wrap up my first WordCamp experience. It is quite a nice meetup for the WordPress users and developers, and the future of WordPress looks pretty good as the number-one open source blogging platform. I definitely learned a lot from discussions with the core developers, but I have to say that not that many attendees of WordCamp are actually developers or contributors, so the talks even on the developer’s day are not really deep into the WP core. Maybe I am just expecting too much here

One suggestion I have is that probably next year, the event should be broadcasted live to the World. After all, many WordPress developers do not have the time and money to attend such an event. Live broadcasting would make things much better.

[END]

As a WordPress plugin developer and also a wp-hacker, I was quite excited about the technical stuff the core WP developer team talked about during the second day.

The first talk(view the slide here) was from Barry and Matt. Barry, the sysadmin, mentioned some amazing stats from WordPress.com and talked about the performance tuning, like APC and WP-CACHE. Matt then introduced HyperDB, the failsafe/redundant/cluster MySQL database class used in the WordPress.com backend. Although I am on the HyperDB maillist since day one, I was having doubts about HyperDB. However, from what Matt talked about, HyperDB does actually many things that I have in mind. Coupled with the new MySQL Proxy, High Availability WordPress clusters would be much easier to build now.

[to be continued…]

The graphical programming tool, Labview, is great for scientific instrumentation, because it is easy to use and yet very powerful and feature rich. However, the graphical interface sometimes can be a hassle, especially when you want to do complicated mathematical operations. Like a look at the example here for adding up two 1’s. Drawing the wires can be a headache if you are doing many many more operations.

This is when you will need the MathScript in Labview. It is similar to MATLAB syntax and does a good job complimenting the all-graphical programming of Labview.

Today, I received an email saying that NI just released an online beta demo (link). With the demo, you can actually run MathScript programs without installing Labview! It can even output 2D or 3D plots. This is great for plotting function for any school project

Try some simple codes:

a= 0 : 0.1 : 2*pi
plot(a, sin(a))

???????????????????????????

Anyway, as I started playing with .hws file a bit more, I found it quite easy to use, and because it is based on HDF5, I can use any software that supports HDF5 to open the HWS file! And, no surprise here, MATLAB, from ver 6.5, has functions to read and write HDF files.

However, MATLAB’s hdf5read() function requires an unique dataset attribute for the data that you want to load from the HDF5 file. So, to read the NI-HWS file, we need to figure out the dataset attribute. Using HDF Explorer and HDF Viewer application, I was able to find the correct attribute to read the HWS file in MATLAB. Here it is

hdf5read('test.hws','/wfm_group0/axes/axis1/data_vector/data');

The second paramter is the HDF5 attribute for the actual dataset in the HWS file.

1968 is actually a very important year in the history of UCSD. The first undergraduate class of UCSD was admitted in 1964, so 1968 was the first class of UCSD graduates! Anyway, the photo below is the campus photo. At that time, only Revelle College was there. In the 40 years after this photo was taken, UCSD has grown to 6 colleges and more than 20000 students. What an accomplishment!

More history of UCSD on TritonLink?

For comparison, you can take a look at the recent satellite and aerial images. Google Maps, and Microsoft Live Local, or the official UCSD Maps.

apt-get install php4-pear php4-dev

In order to use the ‘pecl’ installation tool, you need to upgrade the bundled PEAR version to the latest:

pear upgrade pear-1.3.3
pear upgrade Archive_Tar
pear upgrade Structures_Graph
pear upgrade pear
pear upgrade-all


After the upgrade, you need one more step before proceeding with the pecl installation. If you don’t do the following you will have the error like this when compiling php-memcache: (configure: error: Cannot find php_session.h). Thanks to this post, I found the solution: you need to create the following symoblic link:

ln -s /usr/include/php4 /usr/include/php

Okay, final step is easy:

pecl install apc memcache


For APC, you may just choose not to compile with aspx-support, if it prompted you to select.

Now sit back, and wait for the compilation to finish. After it is done, you just need to enable the modules in your PHP.ini configuration file, which I am not going to touch upon here.

Good luck with your optimization for PHP code.

When trying to build the Ultimate WordPress MU (MultiUser) Appliance, I started to learn rBuilder Online, which I found is not as newbie-friendly as it should be. Thus I wrote this post to record my learning and hopefully it will be shorten your learning time.

The purpose of this tutorial is to get you started in setting up a rBuilder Online project. and use a local rPath Virtual Machine to checkout, test, build and commit your own packages to rBuilder Online. In the end, you will be able to create a virtual machine disk image online, containing your customized Linux.

If you just want to use pre-built packages and organize them into a customized Linux, you can simply use the web interface in rBuilder Online. But if you need to build your own package, please follow the guide here.

To start, you will need at least VMWare Player or VMWare Server (Fusion if you are using Intel Macs) installed on your machine. Make sure you have at least 512MB of physical memory, because the Linux Virtual Machine needs some of your physical memory.Once you have set up your Virtualization environment, download the rPath Linux VMWare Player image. Go to the rPath Linux project website, find a release (currently 1.0.5) and download the VMWare Player image.

While you are downloading the file, get yourself a free account on rPath Builder Online and create a new project. However, please note that the projects hosted on rBuilder Online would be open-source, which means if you want to keep your code secret, you will need to set up your own rBuilder environment. I don’t know how to do yet, but be sure to check out the wiki, though.

All ready! I guess you already got the downloaded zip file for rPath Linux. Unzip it to a folder and use VMWare Player/Server to open the .vmx file. Boot up the Virtual Machine. You will be soon welcomed by a graphical interface from rPath Linux.

Use username: root and empty password to log in.

Open up a text terminal (upper-left corner: Application -> System Tools -> Terminal)

type in the following commands to update the software and install the rmake building tool:

conary updateall
conary update rmake --resolve
service rmake on


Now, the system is almost ready for building packages. But we need to make some new packages. I am not going to the details here because you can follow the easy tutorial in the wiki. Note, you cannot cook new packages using root user, so you will need to create yourself a new user in the system.

After finding the right recipe for your application and cooking it for a test, you are about ready to build it into a trove.

Take a quick look at the workflow for rMake. For a simple package, you can just do this:

rmake build wpmu.recipe

Finally, commit your build into the rBuilder Online repository. Replace the number 1 with any job number you see when you use rmake to build your recipe.

rmake commit 1

Now, you can simply go to rBuilder Online and use the web interface to add your package to a group and cook it.

After cooking the group, you can make a build based on it, and download the disk image with your own customizations! Have fun!

A screenshot of the main index page for the Virtual Appliance.

Image credit: The Boar Year 2007 by ~ysanau on deviantART

The best way I found was to cache your site using the WP-Cache-2 Plugin by Ricardo Galli.

For a blog site, most contents usually won’t change very often. Even if you are really energetic and passionate, one post a day would be probably the most you can do.
However, the comments on your post can change much more frequently. If you set the cache to expire in 1 hour, new post or new comments in that hour would not show up because the old cache is still there. I guess this is still a bug for the plugin, but should be able to get fixed.

Other than the bug, the caching solution worked very well. I used Apache Benchmark (ab) tool to test the speed before and after enabling the caching plugin. The speed boost is almost 4 times with caching (from 800ms/request to 200ms/request)!

If you can bear with the cache expiration bug, this plugin is highly recommended for a busy WP site!