Branch: refs/heads/ehb54-issue-972 Home: https://github.com/ehb54/ultrascan3 Commit: 7a83b64352e047e10fe6bc46d68e4e6a05bd9aee https://github.com/ehb54/ultrascan3/commit/7a83b64352e047e10fe6bc46d68e4e6a0... Author: ehb54 <brookes@uthscsa.edu> Date: 2026-07-19 (Sun, 19 Jul 2026) Changed paths: M us_somo/develop/src/us_hydrodyn_grpy.cpp Log Message: ----------- somo/grpy: fix in-process finish hand-off + resolve input path Two fixes found while validating the in-process path end-to-end via the batch gui_script on a real bead model: 1. grpy_finished() was never invoked. It was posted with QMetaObject::invokeMethod(..., Qt::QueuedConnection, Q_ARG(QProcess:: ExitStatus, ...)), but QProcess::ExitStatus is not a registered queued-connection metatype, so the invoke silently failed ("Unable to handle unregistered datatype 'QProcess::ExitStatus'"). grpy_running then never cleared and the batch's while(grpy_running) processEvents() loop hung forever. Replaced both call sites with QTimer::singleShot(0, this, lambda) which defers to the event loop (still no deep recursion through the model batch) but calls grpy_finished() directly, needing no metatype registration. 2. read_native_file() read grpy_last_processed (a bare filename) relative to the process CWD. The old QProcess set its working directory to get_somo_dir(); resolve the path against get_somo_dir() to match, and guard a missing file with a clear error instead of a silent empty solve. Validated: gui_script batch run on a 165-bead model produces a .grpy_res whose every physical observable (rotational diffusion, sedimentation, intrinsic viscosities, all 8 relaxation times, diagonal diffusion tensor) matches the Fortran GRPY_osx10.11 binary on the same .grpy input exactly; only the documented eigenvector sign-gauge and ~1e-9 coupling noise differ. Refs ehb54/ultrascan-tickets#972 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> To unsubscribe from these emails, change your notification settings at https://github.com/ehb54/ultrascan3/settings/notifications