Author Archive for KwangErn LiewPage 3 of 13

Day 1 FOSDEM Adventure

Published
by
KwangErn Liew
on February 23, 2008
in General
. 1 Comment

FOSDEM is back again!

Tux with Shades, Linux in Hollywood by Robin Rowe and Gabrielle Pantera
This is one of the most internet talk in the event. Both Robin Rowe and Gabrielle Pantera gave insightful information and thoughts on the works of Hollywood studios, for mainly animation/image processing, in Linux. They have introduced a wealth of software, though most are (unfortunately) proprietary software.

Pretty cool to know that movies these days are so realistically created in Linux. Windows sucks? :P

More information can be found at Linux Movies.

One very important aspect they have touched upon is proprietary vs open source software in the industry. Apparently, long time ago, the industry was burnt by the developers, thereby leading them to closed source. Though, another side of the coin is they want to stay competitive in the field by having the best technologies. Open source can stay competitive in the commercial market, but it needs 1) funding/sponsors and 2) excellent support from the community. However, I would somewhat disagree about open source key ingredients (the secret sauce) in creating a successful business; unless you are far ahead in the commercial world and have the critical mass to sustain the application. That’s why I like pluggable software, something like Linux. Individuals can contribute to the core of Linux, where common features and functions are shared, but more specialised needs are pluggable to enhance one’s competitive-ness in the industry.

How a large scale opensource project works by Robert Watson
So, I’ve learnt that FreeBSD is a tremendous project with an amazing history. If only there’s a Debian talk on this, I can objectively compare and contrast both projects to gain a better understanding in large scale opensource projects.

FreeBSD is an interesting project for server needs. However, I’m depressed by the lack of support, albeit (drastically) improving, in XFS and a plethora of software. Even though there are ports and work on the way, it is not yet suited for our needs. Nevertheless, I have obtained a limited copy of FreeBSD 7.0 64bit CD (they were hidden away too!). It’s be interesting to give it a test drive.

Any comments are welcomed about FreeBSD. :)

Linux on the PS3 by Olivier Grisel
How cool is it to have Linux running on PS3?! Even cooler (and cheaper) is to run PS3s as your server farm!! Yellow Dog is the official Linux distribution that fully supports the PS3. I don’t think I saw them there, so no chance in grabbing a free CD. :(

What’s interesting to note is that MIT is running a course in Multicore Programming Primer that looks specifically in programming PS3 Cell Processor. Also, the distributed computing project Folding@home (also known as FAH) is using PS3s to simulate and process research data. Cool stuff!

Think of using PS3 as a processing server. It’s cheap, easily available, and very powerful in today’s standards. Most important question is, can the application utilise its power efficiently? ;) I don’t experimenting if I have the human resource to pump in.

IOGrind: locating I/O performance problems by Michael Meeks
It’s ashame that this is just a lightning talk. IOGrind such an amazing project that I hope it’ll be part of the Linux kernel.

If I understood correctly, the application is able to analyse the disk I/O structure through tracing its (startup) behaviour. In the future, he proposed that when it’s possible to map out the location of all the files to the kernel, there’ll be less seeks on the disk but a direct read/write on the specific inode. Cool stuff. Very interesting indeed.

I hope it gets into the kernel within a couple of years. It should bring on lots of benefits to the Linux server community. Imagine the amount of speed you can improve! It’d definitely be something that I’d want to check out in the future. :)

Day 1 Conclusion
Overall great!

I was also able to check out Openmoko (community site). I proudly crashed it too. Haha :D Must be the old buggy hardware and alpha stage of the software. With the new hardware releasing soon, I don’t think Openmoko will be available to the market any time soon. :( Maybe end of the year? It sure is a cool product and lots of people are interested it in.

Let’s hope next time I touch it, it won’t crash. Haha~

Creating a glossary in TWiki

Published
by
KwangErn Liew
on February 20, 2008
in Development
. 0 Comments

Updated: Click here

TWiki is a great wiki application that has flexibility in creating most types of document management system, let alone project management. Just recently, we require a glossary to ensure we know what we are referring to. Very much similar to what Wikipedia has for Category. TWiki’s FilterPlugin allows such an automated process.

Consider the following in a TWiki topic using TWiki’s text formatting rules:

---+++ A
$ <a name="Application">Application</a> : Of or being a computer program designed for a specific task or use.
---+++ H
$ <a name="Hardware">Hardware</a> : Machines and other physical equipment directly involved in performing an industrial, technological, or military function.
---+++ S
$ <a name="Software">Software</a> : The programs, routines, and symbolic languages that control the functioning of the hardware and direct its operation.

To generate a glossary/index out of it, we must extract the relevant text out of the topic, which FilterPlugin happily does it using %EXTRACT%.

%EXTRACT{ topic="%TOPIC%" expand="off" pattern="^\s{3}\$\s\<a\sname=\"(.*?)\"\>(.*?)\</a\>\s:\s(.*?)[\r\n]+” format=”$1 - $2 :: ” skip=”$3″ }%

What it does is grabbing the content and attribute of the a html tags, and put it in the following format, Application - Application :: Hardware - Hardware :: Software - Software :: . We don’t need the description of the term, so we skip it.

Now, all we need is to put that through %MAKEINDEX% to generate the glossary.

%MAKEINDEX{ "%EXTRACT{ topic="%TOPIC%" expand="off" pattern="^\s{3}\$\s\<a\sname=\"(.*?)\"\>(.*?)\</a\>\s:\s(.*?)[\r\n]+” format=”$1;$2@” }%” split=”@” pattern=”(.*);(.*)” format=”[[#$1][$2]]” }%

For %EXTRACT to work with %MAKEINDEX%, 1) the format of %EXTRACT% output must be slightly changed, 2) let %MAKEINDEX% know where a new definition starts through using split, 3) re-define the pattern of %EXTRACT% output, 4) and define the final output format of %MAKEINDEX%.

That’s it really! You can check out the output here!

Let your imaginations run wild!

Fully Open Source P2P LAMP

Published
by
KwangErn Liew
on November 1, 2007
in Development
. 0 Comments

To build a distributed environment, fully redundant and scalable is not easy and can be expensive! One of the simplest way is to build on top of Amazon’s S3 and EC2. It’s supposedly to be cost effective (pay-as-you-go) and reliable (long awaited release of their SLA). Though I kind of doubt that you can push MySQL onto Amazon’s services.

Now, there’s an alternative to that solution; much easier and more flexible to your own needs.

Kevin Evans (aka kevev) is heading to build a Linux Apache MySQL and PHP/Python/Perl (LAMP) P2P solution naturally called P2PLAMP. The whole idea behind this initiative is to create a clustered version of the Linux OS function as a very robust and fast web server. The nature of its distribution is based on an industrial grade clustered distributed filesystem - GlusterFS - which is managed solely by ZResearch with the support of the Gluster community.

And the best is…its all in the name of Open Source Software. :D