Posts

Craig Latta - Livecoding Language Models: AI Context in Live Environments - 26 March 2025

For our March meeting, Craig Latta will show how to leverage AI language models in a livecoding environment. Using a tool creates context for complimentary tools. For example, selecting an object in a Smalltalk inspector creates useful context for browsing the class of that object. This is a potent idea in an integrated development environment, where multiple tools operate on related information. The extensive context in a live environment enables powerful collaboration with contemporary AI language models. Tools built from that collaboration could greatly augment human capability. Craig Latta will show some foundations for them, and suggest ways of easing their construction through the systematic description of existing tool functionality. For a preview, please see Craig's blog at thiscontext.com . Craig Latta is a research computer scientist in Berkeley and Amsterdam, with interests including livecoding, music performance, and interactive visualization. The discover...

Richard Ronteltap - SmallJS - 26 February 2025

Image
For our February 2025 presentation, Richard Ronteltap will present his SmallJS project. SmallJS is a free and open source implementation of the elegant and powerful Smalltalk-80 (ST) language. It compiles to JavaScript (JS) that runs in modern browsers or in Node.js. SmallJS is file based, not image based, so you can develop in your favorite IDE. The default setup is for Visual Studio Code, with ST syntax coloring and step debugging! You code separately from the SmallJS base libraries (image) and only the parts you use are imported automatically when running your app. SmallJS is fully object oriented, so customizable on every level. For usability, ST class and method names are kept mostly equal to their familiar JS counterparts. JS libraries already encapsulated in ST are: Browsers: Document, Window, HTML elements, events, CSS, streams. Node.js: HTTP server, Express, 4 databases, files. SmallJS is mostly intended for use by JS devs that want a nicer language to work in...

2024 retrospective and what to expect in 2025 - 29 January 2025

For our first meeting of the year, we'll hold a retrospective on what has happened in the Smalltalk world in 2024, and discuss what we can expect in 2025. Join us to share what you learned, and what you are looking forward to! This will be an online meeting from home. If you'd like to join us, please sign up in advance on the meeting's  Meetup page  to receive the meeting details. Don’t forget to bring your laptop and drinks!  

Christoph Thiede on SemanticText and Guille Amaral on Webside - 27 November 2024

Image
Join us next Wednesday for an engaging double presentation on innovative advancements in Smalltalk programming: ### 1. SemanticText: Improving Exploratory Programming in Squeak with Generative AI Presenter: Christoph Thiede From autocomplete to conversational agents, AI is transforming how we interact with code. Christoph will introduce SemanticText , a framework that integrates large language models (LLMs) into Squeak/Smalltalk, enabling conversational agents, semantic search tools, and retrieval-augmented generation workflows. He will showcase experimental integrations with Squeak’s tools and present the concept of a semantic exploratory programming system for debugging and exploring systems using natural language. Bio: Christoph Thiede is a member of HPI’s Software Architecture Group and a core developer of Squeak/Smalltalk, with a focus on enhancing developer productivity and tools. *** ### 2. Webside: A Unified HTTP API for Smalltalk Presenter: Guillermo Amaral ...

On Large Language Models and Smalltalk -- 30 Oct 2024

Image
Meeting Announcement Date: Wednesday, 30th October 2024 Time: 7:00 PM GMT Join us for an engaging session on the latest in Generative AI and Large Language Models (LLMs), featuring two insightful presentations. Introduction to Large Language Models and Gloodata Speaker: Mariano Guerra Mariano will walk us through the fundamentals of LLMs and provide a behind-the-scenes look at building the Gloodata application. A Unique Perspective on LLMs Speaker: Tudor Girba Tudor will offer a distinct view on LLMs, showcasing demos based on GT that highlight their innovative approach. We look forward to your participation! [1] https://gloodata.com [2] https://feenk.com  Update 8 January 2025: The recordings of the presentations are now available on YouTube:  

Christian Haider: How values make systems simpler - 25 Sep 2024 @ 7pm BST

Image
How values make systems simpler Distinguishing between stateful Objects and Values has long been recognized as fruitful. Values are universal context-free abstractions that prevent side-effects and allow for a functional programming style. Though programming languages provide simple Values like Integer and String, more complex Values are usually not supported. We show how to model complex structures as Values in Smalltalk, which allows for a simple and versatile implementation. The resulting Smalltalk systems are simple, clear and easily testable. Christian Haider  is the author of the  PDFtalk  library used in  smallCharts  and other projects. Underlying is a simple implementation of  Values  which he uses extensibly. Christian will show how Values make systems simpler and easier. Examples for some uses and the limits of the approach will be discussed. Since Values use only standard Smalltalk features, it can be used in most other Smalltalks. And all ...

Javier Pimas -- Live Metacircular Runtimes: The case of Egg Smalltalk - 29 June 2024

Image
 Egg is a new Smalltalk dialect that was designed from scratch to incorporate some interesting features: A module system with namespaces that replaces the old-good Smalltalk global. Dynamic identifiers, which are bound lazily similarly to how methods are lazily bound. A multi-VM architecture, with different VM implementations written in C++, Pharo, JavaScript and Egg. The Egg-in-Egg VM is special in that the VM component is just another module of the system, creating what we have named Live Metacircular Runtimes (LMRs) [1]. The most interesting characteristic of LMRs in Smalltalk is that they can be developed using standard Smalltalk tools, which shorten feedback loops when doing VM development. During the talk I'll show a little bit about Egg and its LMR, and how not only VM developers get more productive when writing VMs, but also application developers can better understand what the VM does behind the scenes. [1] https://arxiv.org/abs/2312.16973 - Live Objects All Th...