https://bugs.winehq.org/show_bug.cgi?id=46734
Andrew Nguyen arethusa26@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |NEW Ever confirmed|0 |1 CC| |jacek@codeweavers.com URL| |https://web.archive.org/web | |/20190928104526if_/http://d | |ownload.jgsoft.com/editpad/ | |SetupEditPadProDemo.exe Regression SHA1| |7eb785416193f7ca154872222b9 | |cfcd3db7a7418
--- Comment #6 from Andrew Nguyen arethusa26@gmail.com --- After reproducing in a VM the exact environment described, I performed a regression test, which identified the following commit:
7eb785416193f7ca154872222b9cfcd3db7a7418 is the first bad commit commit 7eb785416193f7ca154872222b9cfcd3db7a7418 Author: Jacek Caban jacek@codeweavers.com Date: Wed Oct 4 15:18:29 2017 +0200
server: Use server side named pipe implementation in byte mode.
Signed-off-by: Jacek Caban jacek@codeweavers.com Signed-off-by: Alexandre Julliard julliard@winehq.org
However, I think the fundamental issue is that the current implementation of named pipes in the server no longer relies on allocating a Unix socket pair. Thus, if a named pipe is allocated to capture the output of a process spawned from Win32, there's no longer an fd that can be mapped to a Unix process's stdout/stderr.
This issue may very well be a WONTFIX if there's not an easy way in the current implementation to emulate one end of a named pipe for use by a Unix process.