sb> Mike Hill <mikehill(a)hgeng.com&gt; writes: sb> You're getting bitten by this code in src/syssignal.h: sb> /* SIGPOLL is the SVR4 signal. Those systems generally define sb> SIGIO as an alias for SIGPOLL, but just in case ... */ sb> #if !defined (SIGIO) && defined (SIGPOLL) sb> # define SIGIO SIGPOLL sb> #endif My patch to Fsignal_process may have caused this breakage. But I looked at the code, and it seems to me that it ought to work, even if SIGIO has been redefined to SIGPOLL. Could someone investigate this (Mike?) ? One useful debugging tool is cd src; make process.i to generate preprocessor output (the .i generation capability is a recent enhancement to the build process). Martin

Andy> The problem is that cygwin does this: Andy> #define SIGIO 23 Andy> #define SIGPOLL SIGIO Andy> cygwin32.h does this Andy> #define BROKEN_SIGIO Andy> SIGIO just undef'ed but not SIGPOLL hence the lossage OK, I understand. So Fsignal_process stays as is, and the problem gets resolved in the cygwin header files? Martin