From the gpt-5.6-sol eval (P0, effort L, multi-session). .gitmodules declares main but the submodule pins b64213c5a on feat/surface-pty-accessors; HEAD is NOT an ancestor of fork origin/main (~1778 main-only vs 15 feature-only commits). The feature lineage carries this session's output-tap/pty-accessor/revival work but MISSES fork main's renderer serialization, bounded screen-export, and GHOSTTY_API/mobile-mirror work. The feature commits modify central C/renderer/termio contracts, so the merge is conflict-heavy (confirmed: 4 conflicting files when attempted). Fork docs also still name an obsolete pin ae3cc5d29 (docs/ghostty-fork.md:15).
First step: rebase/merge the ~4 substantive output-tap/accessor/revival commits onto fork main, run the fork test matrix, push main, update the parent submodule pointer + the fork ledger. High regression risk; own session.
From the gpt-5.6-sol eval (P0, effort L, multi-session).
.gitmodulesdeclaresmainbut the submodule pinsb64213c5aonfeat/surface-pty-accessors; HEAD is NOT an ancestor of forkorigin/main(~1778 main-only vs 15 feature-only commits). The feature lineage carries this session's output-tap/pty-accessor/revival work but MISSES fork main's renderer serialization, bounded screen-export, and GHOSTTY_API/mobile-mirror work. The feature commits modify central C/renderer/termio contracts, so the merge is conflict-heavy (confirmed: 4 conflicting files when attempted). Fork docs also still name an obsolete pinae3cc5d29(docs/ghostty-fork.md:15).First step: rebase/merge the ~4 substantive output-tap/accessor/revival commits onto fork
main, run the fork test matrix, push main, update the parent submodule pointer + the fork ledger. High regression risk; own session.