PATCH: remove unused autoconf tests

PATCH: remove unused autoconf tests

[Oliver Kiddle]

I checked for cases of function and header checks where the outcome of
the test is ignored. Much of this is bits of the signal and NIS handling
that I must have missed. The test for xw looks like something that got
added by accident.

There remain many other things explicitly for NeXT, Reliant Unix/sinix,
BeOS, SCO, sysv4, esix, internix, Cray unicos. Sadly IRIX is also
questionable. Pre 10.6 of MacOS also gets some special handling. Does
anyone regard any of these as still relevant or have evidence that
recent zsh would even work on any of them without a lot of work.

Given that we decided to allow C99, can the variable array macros also
perhaps go?

While rather newer, does the unicode9 stuff still serve a purpose? As I
understand it you needed newer unicode support in your terminal emulator
than your libc and Unicode has now moved on a good few versions. So
there was only ever a small window of systems for which this was
relevant(?) But as my unicode usage rarely stretches much beyond umlauts
and the odd Euro symbol someone else might be better able to extract
this cleanly.

Oliver

dife --git a/Src/signals.h b/Src/signals.h
index ea7e780ac..3c329eb42 100644
--- a/Src/signals.h
+++ b/Src/signals.h
@@ -126,10 +126,5 @@
 
 #define queue_signal_level() queueing_enabled
 
-#ifdef BSD_SIGNALS
-#define signal_block(S) sigblock(S)
-#else
 extern sigset_t signal_block (sigset_t);
-#endif  /* BSD_SIGNALS   */
-
 extern sigset_t signal_unblock (sigset_t);
diff --git a/configure.ac b/configure.ac
index 474f0ccee..7073f4e2c 100644
--- a/configure.ac
+++ b/configure.ac
@@ -634,7 +634,7 @@ if test "x$enable_pcre" = xyes; then
 fi
 
 AC_CHECK_HEADERS(sys/time.h sys/times.h sys/select.h termcap.h termio.h \
-		 termios.h sys/param.h sys/filio.h string.h memory.h \
+		 termios.h sys/param.h sys/filio.h \
 		 limits.h fcntl.h libc.h sys/utsname.h sys/resource.h \
 		 sys/sysctl.h sys/random.h \
 		 locale.h errno.h stdio.h stdarg.h varargs.h stdlib.h \
@@ -849,11 +849,6 @@ esac])
 
 AC_SEARCH_LIBS(getpwnam, nsl)
 
-dnl I am told that told that unicos reqire these for nis_list
-if test `echo $host_os | sed 's/^\(unicos\).*/\1/'` = unicos; then
-  LIBS="-lcraylm -lkrb -lnisdb -lnsl -lrpcsvc $LIBS"
-fi
-
 if test "x$dynamic" = xyes; then
   AC_CHECK_LIB(dl, dlopen)
 fi
@@ -1258,13 +1253,12 @@ AC_CHECK_FUNCS(strftime strptime mktime timelocal \
 	       fpurge fseeko ftello \
 	       mkfifo _mktemp mkstemp \
 	       waitpid wait3 \
-	       sigaction sigblock sighold sigrelse sigsetmask sigprocmask \
 	       sigqueue \
 	       killpg setpgid setpgrp tcsetpgrp tcgetattr nice \
 	       gethostname gethostbyname2 getipnodebyname \
 	       inet_aton inet_pton inet_ntop \
 	       getlogin getpwent getpwnam getpwuid getgrgid getgrnam \
-	       initgroups nis_list \
+	       initgroups \
 	       setuid seteuid setreuid setresuid setsid \
 	       setgid setegid setregid setresgid \
 	       memcpy memmove strstr strerror strtoul \
@@ -1275,7 +1269,7 @@ AC_CHECK_FUNCS(strftime strptime mktime timelocal \
 	       signgam tgamma \
 	       log2 \
 	       scalbn \
-	       putenv getenv setenv unsetenv xw\
+	       putenv getenv setenv unsetenv \
 	       brk sbrk \
 	       pathconf sysconf \
 	       tgetent tigetflag tigetnum tigetstr setupterm initscr resize_term \

Re: PATCH: remove unused autoconf tests

[Lawrence Velázquez]

On Sun, Sep 15, 2024, at 7:45 PM, Oliver Kiddle wrote:
> Pre 10.6 of MacOS also gets some special handling. Does
> anyone regard any of these as still relevant or have evidence that
> recent zsh would even work on any of them without a lot of work.

I maintain zsh for MacPorts, and we do have a handful of users who
continue to use ancient systems.  I goof around on a 10.5 VM every
so often, and zsh seems to work there out of the box, but I don't
exactly stress-test it.

-- 
vq

Re: PATCH: remove unused autoconf tests

[Jun. T]

> 2024/09/16 8:45, Oliver Kiddle <opk@zsh.org> wrote:

> Given that we decided to allow C99, can the variable array macros also
> perhaps go?

I think so, although variable length array is "optional" in C11
(mandatory in C99).

> While rather newer, does the unicode9 stuff still serve a purpose?

In commit 171e7fa I modified configure.ac so that unicode9 is used if
either wcwidth() or iswprint() is broken on macOS. At that time (May 2017),
if I remember correctly, both functions were broken in the latest macOS
(10.12 Sierra) and maybe older versions of macOS also.

Later Apple has fixed the problem but I don't know when; at least on
macOS 12 (Monterey) unicode9 is not required. If we want to support
older macOS then it would be better to keep unicode9. Or we can just
say to users of old macOS "please use zsh-5.9".

> 2024/09/25 7:43, Lawrence Velázquez <larryv@zsh.org> wrote:
> 
> On Sun, Sep 15, 2024, at 7:45 PM, Oliver Kiddle wrote:
>> Pre 10.6 of MacOS also gets some special handling. Does
>> anyone regard any of these as still relevant or have evidence that
>> recent zsh would even work on any of them without a lot of work.
> 
> I maintain zsh for MacPorts, and we do have a handful of users who
> continue to use ancient systems.  I goof around on a 10.5 VM every
> so often, and zsh seems to work there out of the box

Does the complier for 10.5 (old gcc, not clang, I guess) support C99?.