Smalltalk's Browser: Unbeatable, yet Not Enough

5 hours ago

The four-pane System Browser has been central to Smalltalk development for 40 years, providing essential context.
Smalltalk pioneered many modern IDE features like live inspection and powerful navigation decades ago.
The System Browser's strength lies in its ability to maintain context, crucial for understanding message passing within class structures.
Despite its strengths, the System Browser is limited in representing dynamic workflows and message flows across multiple tools.
Smalltalk IDEs suffer from tool integration issues, leading to chaotic workflows and usability friction.
Common problems include 'Frankenstein tools' with accumulated features, 'hermit tools' that don't integrate well, and 'alien tools' that clash with modern workflows.
The growing complexity of Smalltalk systems (e.g., Pharo's 10,750 classes vs. Smalltalk-80's 223) exacerbates navigation and discoverability challenges.
The real issue isn't the System Browser itself but the lack of composition and dynamic context across IDE tools.
A potential solution involves treating the workspace as a graph of related tools organized around investigation threads rather than independent windows.
Future improvements may focus on making the entire IDE navigable, tracking exploration paths and tool relationships, rather than replacing the System Browser.

Hasty Briefsbeta