On Fri Apr 3 22:13:12 2026 +0000, Matteo Bruni wrote:
Very interesting. Can you please add that to the patch as a small comment? It might be nice to mention what's the objective of this "forced sync", why we can do it (i.e. the mask bits are chosen so that the cumulative error to `opos_num` doesn't affect the `opos` and `idx` values computed in the loop) and where does that "23" come from. Maybe it should be obvious, but I'd err on the side of overexplaining rather than the opposite, at least in this case. Also I'd favor a note on the `^ (DWORD)opos_num_mask` part, to me that's not immediately trivial (especially as it will appear after this patch goes in i.e. without the previous form of the `rem` expression available to compare). Done.
-- https://gitlab.winehq.org/wine/wine/-/merge_requests/10423#note_135427