EByte 0.3.0 removes the chore its earlier case studies kept running into: changing the length of anything in a file used to mean fixing a size field somewhere else by hand. A schema can now say where a number comes from (the length of a field, the length of a run of fields, where a field starts, or how many blocks a repeat holds, plus a constant) and the encoder writes it. There’s no general arithmetic. Decoding checks those numbers too, so a file with a stale size fails at that field instead of being quietly corrected on the way back out.
The case study ran the released binary on real samples. Cutting a WAV file’s audio in half took one edit; EByte rewrote the chunk size and riff_size, and ffprobe and Python’s wave both read 0.05 seconds. Renaming the last file in a three-file ZIP updated both name lengths and the directory size and offset, and unzip -t and zipfile were happy. Renaming the first file showed the limit: each entry’s local header offset lives in a different entry, so those two numbers still needed a hand edit before either tool would read the archive. The schema library now derives what it can for WAV, BMP and ZIP.
An independent review found nothing blocking; its small notes were fixed before the release. Binaries for Linux, macOS and Windows are on the latest release, open source under Apache 2.0.