A canvas side was cast to int with no range check, so a number that did
not fit came back as one that did: `-s 4294968296x600` rendered a
perfectly plausible 1000x600, and a hand-edited project file could ask
for a width the reader turned into something else (or, for a double out
of range, into undefined behaviour).
Both front-ends now check the side before the cast, against one shared
bound in chain.hpp — the 1..20000 the GUI spin buttons already offered,
so the help text, the CLI error, the project reader and the interface
all move together.
The tests gain a `canvas size` section covering the round trip at both
ends of the range and the refusals on either side of it, through the
project reader and through `-s`; the test binary picks up the CLI parser
to do it, which stays GUI-free.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Port of the Python prototype to C++20 with a CMake build:
- effector_core: GUI-free static library (effects, image IO via gdk-pixbuf,
declarative param specs, registry, portable seeded RNG, thread helper)
- scanlines effect: plans strips from the seeded RNG, then copies pixels in
parallel with memcpy and doubling horizontal tiling
- gtkmm 4 interface: native file dialogs, widgets generated from the param
specs, coalesced background renders, preview as a GDK memory texture
- CLI: flags generated from the effect registry, headless rendering with no
display required
- ctest suite covering the RNG, registry, effect geometry and image round trip
The Python prototype is kept under python-reference/.
