summaryrefslogtreecommitdiff
path: root/src/s/isc2-2.h
diff options
context:
space:
mode:
Diffstat (limited to 'src/s/isc2-2.h')
-rw-r--r--src/s/isc2-2.h18
1 files changed, 8 insertions, 10 deletions
diff --git a/src/s/isc2-2.h b/src/s/isc2-2.h
index e2f4c988e5e..f669fc2c141 100644
--- a/src/s/isc2-2.h
+++ b/src/s/isc2-2.h
@@ -9,11 +9,14 @@
have been observed in ISC 3.0. */
#define BROKEN_SELECT_NON_X
-/* Although ISC has sockets, again in -linet, again it's not what Emacs
- needs. With this defined, interrupt-shell-subjob and the like do
- nothing. --karl@cs.umb.edu
-#define HAVE_SOCKETS */
+/* karl@cs.umb.edu says that ISC's socket support (in -linet) isn't
+ what Emacs needs; it makes interrupt-shell-subjob and the like do
+ nothing. But that appears to have been another manifestation of
+ the broken select, so it should now be safe to define this again. */
+#define HAVE_SOCKETS
+#define NO_SOCKETS_IN_FILE_SYSTEM
+#define NEED_NET_ERRNO_H
/* This keeps the .cdbx section that gcc puts out when generating
stabs-in-coff output, so Emacs can be debugged. --karl@cs.umb.edu. */
@@ -51,13 +54,8 @@
#define NO_X_DESTROY_DATABASE
/* -linet may be needed to avoid undefined symbols such as gethostname,
- inet_addr, gethostbyname, socket, connect, ... But if we are not
- compiling with X support, it's not needed. */
-#ifdef HAVE_X_WINDOWS
+ inet_addr, gethostbyname, socket, connect, ... */
#define LIBS_SYSTEM -linet LIB_STANDARD_1
-#else
-#define LIBS_SYSTEM LIB_STANDARD_1
-#endif
/* This system has job control. */
#undef NOMULTIPLEJOBS