I recently installed a fresh version of Cygwin on a machine and had my build system fail, due to the previously discussed issue with cp and .exe files. In a nutshell, "touch foo.exe; cp foo bar" fails. See:
http://cygwin.com/ml/cygwin/2003-10/msg00989.html http://cygwin.com/ml/cygwin/2003-09/msg01196.html I tracked it down to the removal of a block of code between fileutils-4.1-1 and fileutils-4.1.2: --- fileutils-4.1-1/./src/copy.c 2001-06-15 15:20:08.000000000 -0400 +++ fileutils-4.1-2/./src/copy.c 2001-06-25 18:42:46.000000000 -0400 @@ -197,25 +197,6 @@ source_desc = open (src_path, O_RDONLY); if (source_desc < 0) { -#ifdef __CYGWIN__ - char *p; - if ((p = strchr (src_path, '\0') - 4) <= src_path || strcasecmp (p, ".exe") != 0) - { - p = alloca (strlen (src_path) + 5); - src_path = strcat (strcpy (p, src_path), ".exe"); - source_desc = open (src_path, O_RDONLY); - if (source_desc >= 0) - { - if ((p = strchr (dst_path, '\0') - 4) <= dst_path || - (p[-1] != '.' && strcasecmp (p, ".exe") != 0 )) - { - p = alloca (strlen (dst_path) + 5); - dst_path = strcat (strcpy (p, dst_path), ".exe"); - } - goto ok; - } - } -#endif /*__CYGWIN__*/ /* If SRC_PATH doesn't exist, then chances are good that the user did something like this `cp --backup foo foo': and foo existed to start with, but copy_internal renamed DST_PATH @@ -229,9 +210,6 @@ return -1; } -#ifdef __CYGWIN__ - ok: -#endif /* These semantics are required for cp. The if-block will be taken in move_mode. */ if (*new_dst) @@ -1163,16 +1141,6 @@ { if (chmod (dst_path, get_dest_mode (x, src_mode))) { -#ifdef __CYGWIN__ - char *p; - if ((p = strchr (dst_path, '\0') - 4) <= src_path || strcasecmp (p, ".exe") != 0) - { - p = alloca (strlen (dst_path) + 5); - (void) strcat (strcpy (p, src_path), ".exe"); - if (chmod (p, src_mode & x->umask_kill) == 0) - goto ok; - } -#endif /*__CYGWIN__*/ error (0, errno, _("setting permissions for %s"), quote (dst_path)); if (x->set_mode || x->require_preserve) return 1; This code was added in 4.1-1 in the first place, with the rationale that it's good for make(1) rules. I found this to be true. I've searched for a comment or rationale for removing this, but I haven't found a Cygwin-specific ChangeLog for fileutils, nor CVS for it, nor even an announcement of 4.1-2 (did it ride in with Cygwin 1.5.0-1 or 1.5.1-1?). Was this an intentional change? If so, what was the intent? The previous behavior is very useful. - Nathan -- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Problem reports: http://cygwin.com/problems.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
