tobez.org

While relational databases represent a valuable and versatile tool, they have their problems. For many workloads the performance offered by many RDBMSes does not cut it.

This talk considers one such case, and walks through the implementation of one possible solution, namely the replacement of a PostgreSQL instance with an in-memory key-value store Redis.

The general steps necessary for such conversion are defined and elaborated upon.

The limitations of using Redis instead of an RDBMS are discussed as well.

As a bonus, using Redis as a message bus is described. An implementation of a generic queue for watching Best Practical's Request Tracker's transactions is used as an illustration.

Go to presentation.

The DBIx::Perlish module allows one to use a domain-specific declarative language with Perl syntax instead of SQL for making database queries. This ability comes at a price, however - the parsing of Perl opcode tree that the module performs takes a surprizingly large amount of time.

The talk delves into the implementation details of DBIx::Perlish and its helper modules and then proceeds to discuss challenges involved in making the module cache the generated SQL, together with the solutions to those challenges.

While most of the advanced concepts used by the module will be briefly introduced in the talk, it will help if the audience has at least some level of familiarity with perl optrees.

Go to presentation.

This talk is about the FreeBSD ports collection, with the emphasis on practical things you need to know should you decide to make a port of some software yourself. A number of "how" and some "whys" are explained.

Three ports are created during the session as an illustration.

Go to presentation.

Fetch a tarball with the complete presentation.

Sometimes you want to get an existing sub or a closure and have it stored on disk or in the database, and execute it at some later stage. This talk discusses why you might want to do that, and how you might proceed doing it. The solution involves using one of the more underappreciated CPAN modules out there.

Go to presentation.

Fetch a tarball with the complete presentation.

The DBIx::Perlish module provides the ability to work with databases supported by the DBI module using Perl's own syntax for four most common operations: SELECT, UPDATE, DELETE, and INSERT.

By using DBIx::Perlish, you can write most of your database queries using a domain-specific language with Perl syntax. Since a Perl programmer knows Perl by definition, and might not know SQL to the same degree, this approach generally leads to a more comprehensible and maintainable code.

Go to presentation.

Fetch a tarball with the complete presentation.

Older version from NPW2007, April 29th 2007, Copenhagen, Denmark

Go to presentation.

Fetch a tarball with the complete presentation.

This intermediate-level talk is aimed at people who already know Perl but are not very familiar with its OO model. It discusses at length the standard language constructs provided by Perl that facilitate object-oriented programming, its advantages and deficiencies. The methods are described to work around some of the limitations of the object model. Some more advanced topics include discussing multiple inheritance, building automatic properties accessors and creating singleton classes. Finally, non-idiomatic ways of doing OO in Perl are described, in partilular objects as closures with state, inside-out objects, and objects with dynamic methods via the Class::Prototyped module.

Go to presentation.

Fetch a tarball with the complete presentation.

Perl6 is an ongoing effort by the Perl community to design a dynamic programming language following the principles established by Perl5, but substantially better. Parrot is the virtual machine on which Perl6 (and other languages) will run. Pugs is the initial implementation of Perl6. This talk aims to describe, in broad strokes, the design, the philosophy, and the features of all three.

Go to presentation.

Fetch a tarball with the complete presentation.