Approach
The constraint
is the method.
The work is built on a 2007 dual-core Pentium T3200. Each consumer is chosen not as a showcase, but for the pressure it can apply to the runtime.
Fix it at the root.
Through pre-1.0 the source stayed private on purpose: bugs got fixed at the root instead of papered over for downstream users, with no obligation to preserve old behavior. Every issue was corrected where it lived, the API tightened, and the language got cleaner with each version.
That season is historical.
EigenScript is public. Read the source, build it yourself, and run the same binary that serves this page.
The feedback loop
Workloads are the question.
The method is a loop: point a demanding workload at the runtime, and it turns up a real bug in a signature feature — which gets fixed upstream, at the root, with a regression test.
RED-TEAMED ITS OWN SERVER
The binary serving this site was attacked and revealed seven HTTP availability bugs the test suite missed, including a slowloris route to taking it offline. The accept loop is now multi-threaded with a connection cap; every finding shipped with a regression test.
A SOLVER THAT WOULD NOT CONVERGE
A dynamical-systems lab reported equilibrium on solvers at rest. Observer history had been stored on recyclable, reference-shared objects. Re-keying observer state to the variable binding let those loops converge.
DRIVEN BY THE WORKLOAD
Performance changes trace back to a real consumer hot path: Tidepool’s RL loop or the DMG emulator’s inner cycle. The language gets faster because a program needs it to, measured against a baseline.
THE WHOLE STACK
Language primitives, a JIT, self-training AI, a SAT solver, an emulator, deterministic simulation, and persistent storage are not separate claims. They are adjacent tests of the same system.