Обсуждение: [HACKERS] NEXTSTEP compilation fixes, misc.

Find included below a few patches to help compile the 970608 snapshot with NEXTSTEP.

Umm, is there anybody out there who could explain to me the hidden system
behind the portability directories ? I.e. NEXTSTEP has no strdup() and no
putenv. There's one strdup() implementation in
src/backend/port/nextstep/port.c, another one in utils/strdup.c, a third one
(pstrdup) in backend/utils/mmgr/palloc.c. putenv gets defined in
src/backend/port/nextstep/port.c, but is therefore not available e.g. in
src/bin/psql. This looks like rather uncoordinated hacks all over the place.

Wouldn't it be more reasonable to have only one utility library that provides
all the functions needed for the specific system ? My solution was to hack
bin/psql/Makefile to include backend/port/SUBSYS.o instead of the STRDUP
conditional. No idea if this breaks anything else.

    Gregor


- ---
| Gregor Hoffleit                        Mathematisches Institut, Uni HD |
| flight@mathi.uni-heidelberg.de      INF 288, 69120 Heidelberg, Germany |
| (NeXTmail, MIME)                      (49)6221 54-5771    fax  54-8312 |
| PGP Key fingerprint = 23 8F B3 38 A3 39 A6 01  5B 99 91 D6 F2 AC CD C7 |


- --- src/bin/psql/Makefile.in.ORIG    Sun Jun  8 15:23:06 1997
+++ src/bin/psql/Makefile.in    Sun Jun  8 15:25:56 1997
@@ -16,15 +16,12 @@

 CFLAGS+= -I$(LIBPQDIR)

- -OBJS= psql.o stringutils.o @STRDUP@
+OBJS= psql.o stringutils.o

 all: submake psql

 psql: $(OBJS) $(LIBPQDIR)/libpq.a
- -    $(CC) -o psql -L$(LIBPQDIR) $(OBJS) -lpq $(LDFLAGS)
- -
- -../../utils/strdup.o:
- -    $(MAKE) -C ../../utils strdup.o
+    $(CC) -o psql -L$(LIBPQDIR) ../../backend/port/SUBSYS.o $(OBJS) -lpq
$(LDFLAGS)

 .PHONY: submake
 submake:


- --- src/makefiles/Makefile.nextstep.ORIG    Thu Jan  1 01:00:00 1970
+++ src/makefiles/Makefile.nextstep    Sun Jun  8 13:58:52 1997
@@ -0,0 +1,1 @@
+

- --- src/include/port/nextstep.h.ORIG    Sun Jun  8 14:02:20 1997
+++ src/include/port/nextstep.h    Sun Jun  8 14:02:11 1997
@@ -16,3 +16,16 @@
 # define JMP_BUF
 # define NO_WAITPID
   typedef struct mutex slock_t;
+
+# ifndef BYTE_ORDER
+#  ifdef __BIG_ENDIAN__
+#   define BYTE_ORDER BIG_ENDIAN
+#  else
+#   define BYTE_ORDER LITTLE_ENDIAN
+#  endif
+# endif
+
+# undef FALSE
+# define FALSE 0
+
+# undef HAVE_TERMIOS_H

- --- src/libpq/fe-connect.c.ORIG    Sun Jun  8 14:49:16 1997
+++ src/libpq/fe-connect.c    Sun Jun  8 14:50:08 1997
@@ -21,6 +21,10 @@
 #include <ctype.h>
 #include <string.h>
 #include <netdb.h>
+#if defined(NeXT)
+#include <netinet/in_systm.h>
+#include <netinet/in.h>
+#endif
 #include <netinet/tcp.h>
 #include <errno.h>
 #include <signal.h>

- --- src/backend/libpq/pqcomm.c.ORIG    Sun Jun  8 14:13:18 1997
+++ src/backend/libpq/pqcomm.c    Sun Jun  8 14:13:48 1997
@@ -43,6 +43,9 @@
 #include <sys/types.h>
 #include <sys/socket.h>
 #include <netdb.h>
+#if defined(NeXT)
+#include <netinet/in_systm.h>
+#endif
 #include <netinet/in.h>
 #include <netinet/tcp.h>
 #include <arpa/inet.h>

------------------------------

Gregor Hoffleit wrote:
>
> Find included below a few patches to help compile the 970608 snapshot with NEXTSTEP.

Gregor, does abstime and datetime work correctly under NEXTSTEP? I'd
like to know if NEXTSTEP has POSIX-compatible time zone information
available in library calls. It is one of the only systems supported by
Postgres which does not currently have USE_POSIX_TIME enabled, but that
may not be necessary.

Can you try compiling with USE_POSIX_TIME set in include/config.h?

Thanks.

            - Tom

------------------------------