A Russian programmer's refusal to accept bloated binaries produced one of the most studied alternative frameworks in the Object Pascal world.
The Objection That Drove the Design
By the late 1990s, a minimal Delphi application compiled against the VCL (Visual Component Library) carried a runtime footprint that struck a segment of the community as structurally indefensible. The VCL was a deep class hierarchy built around RTTI, published properties and the streaming system those features required; even a form with a single button dragged in metadata, registration tables and ancestor chains the application had never asked for. Vladimir Kladov, a Russian developer working in the Object Pascal ecosystem, made this overhead the explicit subject of his design work. His answer was the Key Objects Library, known universally as KOL.
KOL's stated premise was that a complete, functional Windows application need not exceed thirty kilobytes. That figure was not approximate or aspirational — it was the benchmark against which Kladov measured every architectural choice. The library dispensed with Delphi's class hierarchy at the framework level, replacing TObject-rooted component trees with records and procedure variables. There were no published properties in the VCL sense, no RTTI tables attached to controls, and no DFM streaming machinery to carry. Windows messages were handled directly. The result was a library that exposed the Win32 API through a thin, stable Object Pascal surface without the abstraction cost the VCL had accepted as necessary.
MCK and the Design-Time Bridge
The companion tool, Mirror Classes Kit (MCK), addressed the practical objection that abandoning the VCL also meant abandoning the Delphi form designer. MCK provided a parallel set of mirror components — VCL-based at design time only — that mapped to KOL objects in the generated unit. A developer laid out a form inside the Delphi IDE as normal; MCK translated the design into KOL-compatible Pascal code rather than DFM properties and VCL component instances. At runtime, nothing from the mirror layer survived into the binary. The designer experience was preserved; the footprint penalty was not incurred.
This two-layer approach made KOL and MCK together a coherent response to a specific professional problem. Embedded developers, shareware authors distributing over slow connections, and anyone targeting low-resource Windows machines had a documented reason to prefer a thirty-kilobyte executable over one an order of magnitude larger. The libraries were released as open source, and Kladov maintained them across multiple Delphi versions, tracking the compiler's evolution and accommodating both the 32-bit and, in later iterations, 64-bit targets that Free Pascal compatibility demanded.
A working records desk of the kind Delphi was bought to build software for — the business applications that made the compiler its market.
Photo: Pavel Danilyuk / Pexels
The Community Record
Delphi-era developer conferences were where version changes got argued out in front of the people who had to migrate.
Photo: Matheus Bertelli / Pexels
The KOL and MCK libraries attracted a stable, technically engaged user base documented in forum archives, dedicated mailing lists and a body of contributed extensions. Developers added database access layers, networking wrappers and additional control sets, all built to the same constraint: no RTTI, no VCL dependency, no hidden size cost. The JEDI community and other third-party contributors produced compatibility code, and Kladov's own documentation — published in Russian and in English — gave a precise account of the design rationale.
What the record shows is not simply a smaller library but a coherent technical argument. The VCL's architecture represented deliberate choices about abstraction, designer integration and component reuse that served the mainstream Delphi use case well. Kladov's work demonstrated that a different set of choices, made consistently and documented honestly, produced a different but legitimate result. The thirty-kilobyte figure was, in this sense, both a measurement and a thesis: a claim that the overhead developers accepted as fixed was in fact contingent on architectural assumptions that could be refused.
Kladov's position in the Object Pascal history is that of a specialist dissenter whose dissent was rigorous enough to attract a community of its own — a technical objection held together by working code.
More in People