2.1.95 result - II - make check (failed, patch attached)

[Matthew Clarke <Matthew_Clarke at mindlink dot bc dot ca>]

I wrote:

[ snip ]

> "make check" running now.  I'll report on that when it finishes.

Failed:

/bin/sh -e tst-printf.sh /build/glibc22/ '/build/glibc22/elf/ld-linux.so.2 --library-path /build/glibc22:/build/glibc22/math:/build/glibc22/elf:/build/glibc22/dlfcn:/build/glibc22/nss:/build/glibc22/nis:/build/glibc22/rt:/build/glibc22/resolv:/build/glibc22/crypt:/build/glibc22/linuxthreads'
tst-printf.sh[256]: >&/dev/null : illegal file descriptor name
tst-printf.sh[256]: >&/dev/null : illegal file descriptor name
*** output comparison failed
make[2]: *** [/build/glibc22/stdio-common/tst-printf.out] Erreur 1
make[2]: Quitte le répertoire `/home/clamat/src/glibc-2.1.95/stdio-common'
make[1]: *** [stdio-common/tests] Erreur 2
make[1]: Quitte le répertoire `/home/clamat/src/glibc-2.1.95'
make: *** [check] Erreur 2

Eh?  From ~/src/glibc-2.1.95/stdio-common/tst-printf.sh, line 252:

	cmp - ${common_objpfx}stdio-common/tst-printf.out >& /dev/null ||

Hrm.  That use of '>&' looks like a bash-ism.  (My /bin/sh is PD KSH.)
Trivial patch attached, for both occurrences of this >& construct.  A
quick fgrep for '>&' on all files ending in ".sh" didn't show any other
trouble like this.

I also note in passing the following compiler warnings in tst-swprintf.c
in stdio-common (copy'n'pasted), in case anyone's interested:

tst-swprintf.c:45: warning: long int format, different type arg (arg 3)
tst-swprintf.c:45: warning: long int format, different type arg (arg 3)
tst-swprintf.c:46: warning: long int format, different type arg (arg 3)
tst-swprintf.c:46: warning: long int format, different type arg (arg 3)
tst-swprintf.c:47: warning: long int format, different type arg (arg 3)
tst-swprintf.c:47: warning: long int format, different type arg (arg 3)
tst-swprintf.c:48: warning: long int format, different type arg (arg 3)
tst-swprintf.c:48: warning: long int format, different type arg (arg 3)
tst-swprintf.c:49: warning: long int format, different type arg (arg 3)
tst-swprintf.c:49: warning: long int format, different type arg (arg 3)
tst-swprintf.c:50: warning: long int format, different type arg (arg 3)
tst-swprintf.c:50: warning: long int format, different type arg (arg 3)
tst-swprintf.c:51: warning: long int format, different type arg (arg 3)
tst-swprintf.c:51: warning: long int format, different type arg (arg 3)
tst-swprintf.c:52: warning: long int format, different type arg (arg 3)
tst-swprintf.c:52: warning: long int format, different type arg (arg 3)
tst-swprintf.c:53: warning: long int format, different type arg (arg 3)
tst-swprintf.c:53: warning: long int format, different type arg (arg 3)

Running new "make check" with patched stdio-common/tst-printf.sh.

Will report later.

Matt.
-- 
>Almost any animal is capable learning a stimulus/response association,
>given enough repetition.
Experimental observation suggests that this isn't true if double-clicking
is involved.			-- Lionel, Malcolm Ray, in a.s.r.
--- stdio-common/tst-printf.sh	Sun Oct  8 17:05:05 2000
+++ stdio-common/tst-printf.sh.new	Mon Oct  9 23:54:50 2000
@@ -138,7 +138,7 @@
 printf ("%hu", 65537) = 1
 --- Should be no further output. ---
 EOF
-cmp - ${common_objpfx}stdio-common/tst-printf.out >& /dev/null ||
+cmp - ${common_objpfx}stdio-common/tst-printf.out > /dev/null 2>&1 ||
 cat <<'EOF' |
 %.4x:	`0012'
 %04x:	`0012'
@@ -249,7 +249,7 @@
 printf ("%hu", 65537) = 1
 --- Should be no further output. ---
 EOF
-cmp - ${common_objpfx}stdio-common/tst-printf.out >& /dev/null ||
+cmp - ${common_objpfx}stdio-common/tst-printf.out > /dev/null 2>&1 ||
 {
   status=1
   echo "*** output comparison failed"