Re: "make check" failure with sort-compress on OSF and Linux

[Paul Eggert]

Jim Meyering <address@hidden> writes:

> my jaw hit the floor when I saw the 500+ gzip zombies
> left hanging while I'm at a breakpoint in merge.

Here's a patch to disable compression by default, so that we can get a
more-stable version out the door.  It also removes the findprog
module, which should be one less thing to worry about porting.

2007-02-24  Paul Eggert  <address@hidden>

        * NEWS: sort no longer compress temporaries by default.
        * bootstrap.conf: Remove findprog.
        * doc/coreutils.texi (sort invocation): The default is to not
        compress.  Don't treat "" specially.
        * src/sort.c: Don't include findprog.h.
        (create_temp): Compress only if the user specified --compress-program.
        * tests/misc/sort-compress: Adjusts tests to match new behavior.

diff --git a/NEWS b/NEWS
index a099877..96ebd82 100644
--- a/NEWS
+++ b/NEWS
@@ -44,10 +44,9 @@ GNU coreutils NEWS                                    -*- 
outline -*-

 ** New features

-  By default, sort usually compresses each temporary file it writes.
+  sort's new --compress-program=PROG option specifies a compression
+  program to use when writing and reading temporary files.
   This can help save both time and disk space when sorting large inputs.
-  The default compression program is gzip, but this can be overridden
-  with sort's new --compress-program=PROG option.

 ** New features

diff --git a/bootstrap.conf b/bootstrap.conf
index e94478a..c54b035 100644
--- a/bootstrap.conf
+++ b/bootstrap.conf
@@ -44,7 +44,7 @@ gnulib_modules="
        config-h configmake
        closeout cycle-check d-ino d-type diacrit dirfd dirname dup2
        error euidaccess exclude exitfail fchdir fcntl fcntl-safer fdl
-       file-type fileblocks filemode filenamecat findprog fnmatch-gnu
+       file-type fileblocks filemode filenamecat fnmatch-gnu
        fopen-safer
        fprintftime fsusage ftruncate fts getdate getgroups gethrxtime
        getline getloadavg getndelim2 getopt getpagesize getpass-gnu
diff --git a/doc/coreutils.texi b/doc/coreutils.texi
index 04c1b4e..99412e4 100644
--- a/doc/coreutils.texi
+++ b/doc/coreutils.texi
@@ -3634,9 +3634,7 @@ Other options are:
 @table @samp

 @item address@hidden
-If @var{prog} is not the empty string, compress any temporary files
-with the program @var{prog} rather than with the default compression
-method.  The default is currently @command{gzip} but this may change.
+Compress any temporary files with the program @var{prog}.

 With no arguments, @var{prog} must compress standard input to standard
 output, and when given the @option{-d} option it must decompress
@@ -3647,9 +3645,6 @@ Terminate with an error if @var{prog} exits with nonzero 
status.
 Whitespace and the backslash character should not appear in
 @var{prog}; they are reserved for future use.

-If @var{prog} is the empty string, do not compress temporary
-files.
-
 @item -k @var{pos1}[,@var{pos2}]
 @itemx address@hidden,@var{pos2}]
 @opindex -k
diff --git a/src/sort.c b/src/sort.c
index 6a7de9c..58ca66a 100644
--- a/src/sort.c
+++ b/src/sort.c
@@ -30,7 +30,6 @@
 #include "system.h"
 #include "argmatch.h"
 #include "error.h"
-#include "findprog.h"
 #include "hard-locale.h"
 #include "hash.h"
 #include "inttostr.h"
@@ -847,14 +846,7 @@ create_temp (FILE **pfp, pid_t *ppid)
   struct tempnode *node = create_temp_file (&tempfd);
   char *name = node->name;

-  if (! compress_program)
-    {
-      static char const default_compress_program[] = "gzip";
-      char const *prog = find_in_path (default_compress_program);
-      compress_program = (prog == default_compress_program ? "" : prog);
-    }
-
-  if (*compress_program)
+  if (compress_program)
     {
       int pipefds[2];

@@ -875,8 +867,7 @@ create_temp (FILE **pfp, pid_t *ppid)
          dup2_or_die (pipefds[0], STDIN_FILENO);
          close (pipefds[0]);

-         if (execlp (compress_program, compress_program,
-                     (char *) NULL) < 0)
+         if (execlp (compress_program, compress_program, (char *) NULL) < 0)
            error (SORT_FAILURE, errno, _("couldn't execute %s"),
                   compress_program);
        }
@@ -925,8 +916,7 @@ open_temp (const char *name, pid_t pid)
       dup2_or_die (pipefds[1], STDOUT_FILENO);
       close (pipefds[1]);

-      if (execlp (compress_program, compress_program,
-                 "-d", (char *) NULL) < 0)
+      if (execlp (compress_program, compress_program, "-d", (char *) NULL) < 0)
        error (SORT_FAILURE, errno, _("couldn't execute %s -d"),
               compress_program);
     }
diff --git a/tests/misc/sort-compress b/tests/misc/sort-compress
index b0f4dd7..0cafb2e 100755
--- a/tests/misc/sort-compress
+++ b/tests/misc/sort-compress
@@ -57,14 +57,14 @@ EOF
 chmod +x gzip

 # This will find our new gzip in PATH
-PATH=.:$PATH sort -S 1k in > out || fail=1
+PATH=.:$PATH sort -S 1k --compress-program=gzip in > out || fail=1
 cmp exp out || fail=1
 test $fail = 1 && diff out exp 2> /dev/null
 test -f ok || fail=1
 rm -f ok

-# This is to make sure we can disable compression
-PATH=.:$PATH sort --compress-program= -S 1k in > out || fail=1
+# This is to make sure it works with no compression.
+PATH=.:$PATH sort -S 1k in > out || fail=1
 cmp exp out || fail=1
 test $fail = 1 && diff out exp 2> /dev/null
 test -f ok && fail=1
M ChangeLog
M NEWS
M bootstrap.conf
M doc/coreutils.texi
M src/sort.c
M tests/misc/sort-compress
Committed as be58204c66dc36ea36fe93b4542e1b664cc4f172