The sessions from this year's TPRC (formerly known as YAPC::NA), the premier Perl developer community conference,
are now available online, for free.

This year's TPRC ran in Las Vegas on June 24-28. With five days packed with great sessions, the real problem was choosing what to miss. No worries though as you can now watch on demand any you have missed, or all of it if haven't attended at all. Given the plethora of material we did a bit of digging to suggest the best to watch first.

Since AI and Coding assistants are a hot subject nowadays, let's start with "PerlGPT, A Code Llama LLM Fine-Tuned For Perl" by William N. Braswell.

PerlGPT is a large language model (LLM) based off Meta's Code Llama that was trained on Perl code so that it can automatically read and write Perl source code
to become a Perl programmer's sidekick. Although the project is still under development, in the session William demonstrated the capabilities of the most up-to-date version of it.

IMHO, there's another powerful model that could potentially be fit as the backend of PerlGPT. As we've examined in "IBM Launches The Granite Code LLM Series", the Granite collection has been trained on code written in 116 programming languages, including Perl, ranging in size from 3 to 34 billion parameters. As a matter of fact in the MultiPL-E: Multilingual Code Generation benchmark, it scored better for Perl against Code Lama.

So just something to keep in mind.

Still on the subject of AI, another insightful session was "Structure Based Structuring of Unstructured Data" by Adam Russel.

A tongue twister but in essence it was about how to apply natural language understanding to information found on the web, something better known as the Semantic Web
which combines format/display with meaning. The prime example is of course Semantic HTML5, which Standard's Draft clearly defines the distinction between display and meaning:

The <section> element is not a generic container element. When an element is needed only for styling purposes or as a convenience for scripting, authors are encouraged to use the <div> element instead. A general rule is that the <section> element is appropriate only if the element’s contents would be listed explicitly in the document’s outline.

This coupled with other such tags which add meaning to your html document, like <article>, <nav> and <address>.

The session goes through such semantics by using the RDF format which subsequently turns into graph for a much better visual experience.

Manually creating semantic graphs of semi/unstructured data such as financial documents is a labor intensive job, so what are the alternatives? Employ LLMs of course to automate the process! The session does that by utilizing various Perl modules; who said that Perl can't do AI like Python does?

Actually Portable Perl was another good session. While Perl is ubiquitous, aka installed on every OS out there, packing Perl executables is not.

Till now the common way of packing an executable resulting from your Perl code, was using PAR::Packer. This in reality would not create a native exe but would instead create a distribution with your source files, the cpan modules they refer to and bundling a Perl interpreter, all in one zip like file, simulating an executable.

While in most cases PAR::Packer would get the job done, in cross-platform scenarios and especially on Windows machines, it came with its own set of issues;just do a search up on Perlmonks to be presented with myriad of threads on such issues. So what is the alternative? Actually Portable Perl (APPerl).

Actually Portable Perl (APPerl) is a distribution of Perl that runs on several x86_64 operating systems via the same binary. The concept is the same as Packer but "Actually Portable" means that it creates an executable which runs on every major platform. Gavin Hayes goes through the process.

Finally I want to conclude by pointing out, "The Once and Future Perl" session done by whom else than Damian Conway. In typical Conway style, the sessions was a mix of philosophy, fun and practicality, talking about the advancements in Perl's type system with a particular focus on subroutine signatures as well as narrating the good and bad parts of applying AI to your Perl code.

There's many outstanding talks still left, so take your time and delve into the conference by going through its official Youtube playlist. Enjoy! 

More Information

TPRC 2024 - Las Vegas, Nevada

Related Articles

Perl v5.40.0 Shows That It Is Too Resilient To Die

IBM Launches The Granite Code LLM Series

Granite Code Models: A Family of Open Foundation Models for Code Intelligence

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info