Example gallery

Each of the following ships under pyccolo/examples/ as a self-contained, tested tracer — great starting points to adapt.

Example	Demonstrates
coverage.py	statement-level code coverage (before_stmt, should_instrument_file)
optional_chaining.py	?., .?, ?? optional chaining / nullish coalescing via AugmentationSpec
pipeline_tracer.py	|> / |>> pipeline operators (binop augmentation)
quick_lambda.py	MacroPy-style f[_ + _] quick lambdas
quasiquote.py	MacroPy-style q[...] / u[...] quasiquotes
block_lambda.py, func_block.py, brace_subscript.py	statement-bodied name{ ... } blocks (paired-delimiter augmentation)
lazy_imports.py	make (most) imports lazy, resolving on first use
future_tracer.py	implicit async: run assignments on a thread pool, unwrap futures on use
concolic.py	concolic (concrete + symbolic) execution with a Z3 / brute-force solver

In the wild

Beyond the bundled examples, Pyccolo is the instrumentation engine behind several larger projects — good places to see what it can do at scale:

• ipyflow — a reactive Python kernel for Jupyter that tracks dataflow between cells using Pyccolo’s dynamic analysis.

• pipescript — a pipe operator (|>), placeholder ($), and macro syntax for IPython/Jupyter, built entirely on Pyccolo’s syntax augmentation and composable event handlers.

• pycograd — a small reverse-mode automatic-differentiation library that differentiates ordinary numpy code (no special “autodiff namespace”), using Pyccolo to trace the computation.

People have also built statement-level coverage, syntactic macros (quasiquotes, quick lambdas), syntax-augmented Python (optional chaining, pipeline operators), lazy imports, concolic execution, and tools to uncover semantic memory leaks.