On Tru64, using a vendor-supplied C compiler, ...
urtx# sizer -v
HP Tru64 UNIX V5.1B (Rev. 2650); Fri Mar 20 20:19:48 CDT 2009
urtx# cc -V
Compaq C V6.5-303 (dtk) on HP Tru64 UNIX V5.1B (Rev. 2650)
Compiler Driver V6.5-302 (dtk) cc Driver
New problems (since 1.5):
urtx# CC=cc ./configure
[...]
urtx# gmake
[...]
CC getcwd-lgpl.o
cc: Warning: getcwd-lgpl.c, line 59: In this statement, "malloc(...)" of type "i
nt", is being converted to "pointer to char". (cvtdiftypes)
buf = malloc (size);
------^
cc: Warning: getcwd-lgpl.c, line 97: In this statement, "realloc(...)" of type "
int", is being converted to "pointer to char". (cvtdiftypes)
ptr = realloc (buf, size);
------^
cc: Warning: getcwd-lgpl.c, line 118: In this statement, "realloc(...)" of type
"int", is being converted to "pointer to char". (cvtdiftypes)
result = realloc (buf, strlen (buf) + 1);
------^
In an environment where "int" is 32 bit and a pointer is 64, this
looked dangerous. I haven't tried to determine which header file
inclusion changed how, but "man malloc" suggests <stdlib.h>, so I threw
one into lib/getcwd-lgpl.c:
urtx# gdiff -u lib/getcwd-lgpl.c_orig lib/getcwd-lgpl.c
--- lib/getcwd-lgpl.c_orig 2013-06-09 15:37:56 -0500
+++ lib/getcwd-lgpl.c 2013-06-10 16:41:52 -0500
@@ -18,6 +18,9 @@
/* Specification */
#include <unistd.h>
+#ifdef HAVE_STDLIB_H
+# include <stdlib.h>
+#endif
#include <errno.h>
#include <string.h>
New "make check" failure:
FAIL: zgrep-context
[...]
+ zgrep -15 17 -
grep: illegal option -- 1
grep: illegal option -- 5
[...]
Same old problems:
CC util.o
cc: Warning: util.c, line 42: Allowing the declaration of a static array with an
incomplete type is a language extension. (incompstatarr)
static const ulg crc_32_tab[]; /* crc table, defined below */
-----------------^
I reported a similar complaint (from a similar compiler) on VMS for
1.5. On VMS, a compiler warning is normally fatal for the build
procedure, so it was more important to fix it there (by moving the
actual array declaration up to where this incomplete declaration was,
and removing this one).
"make check":
FAIL: help-version
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
On VMS, the same set of changes suggested for 1.5 seems to work as
well for 1.6. If it would do more good than it did last year, then I
could regenerate a set of "diff -u" differences for these changes in
1.6.
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
If you have any questions or suggestions (or complaints), please let
me know.
------------------------------------------------------------------------
Steven M. Schweda sms@antinode-info
382 South Warwick Street (+1) 651-699-9818
Saint Paul MN 55105-2547
