A week ago, I just gave a speech about Unix Systems Programming, Chapter 3 - Working with files for the Open Source Community at work. It was practically a two-hour technical discussion on how a Unix-like system sees/handles files and a general overview of syscalls, user-level calls and other topics like the multipurpose fcnt syscall and memory mapped I/O. It was quite fun.

The other side of the story is about Net::RMI, my pure-Perl Remote Method Invocation implementation. Yes, the project is going quite slow, but the design and implementation is near to its final release. While coding the first version of the API, I discovered some design flaws that made me rethink the way the server part was doing its job and also rethink the request-response interchange between the proxy object and the server object.

Here's the UML diagram for the second draft:

As you can see, there are several changes to the API compared to the new version, new classes and a whole new idea on how stuff works internally, In the end, for the user it needs to be transparent.

There's a bad symbol usage in the diagram, the prmiregistry, since this is a class diagram and prmiregistry is not a class, is an independent daemon that waits for requests from the Net::RMI::Registry class. But I wanted to use it to remember that Net::RMI::Registry depends on that process.

Well, we will see what happens, I really hope to finish this stuff as fast as I can, provide a complete test suite and some code examples for the public release.

Until then, have fun.

The current status of Net::RMI is actually good, I'm writting the Net::RMI::Registry and the prmiregistry process server, which will handle all the object activation and registration.

Anyways... yesterday while reading Perl en Español I found a post about a small Perl contest. This contest tests your parsing skills ( in a small scale ), what you need to do is to parse a certain file with a given format, in this case a .srt file, which is a movie subtitiles file format.

The program accepts a filename ( the srt file ) and an amount of time to be added/substracted to each dialog. The fun part is that you cannot use any of the subtitle parsing modules from CPAN and that you need to produce a short/clean code. The output will be sent to STDOUT with the same format as the input file with each dialog modified.

For Spanish readers, here's the post, you can read all the details from there.

Here's my solution (the post has also my solution).

Nice eh? well, laziness is not always that bad, Have fun!.