CVS commit: pkgsrc/sysutils/coreutils

["Makoto Fujiwara" <mef%netbsd.org@localhost>]

Module Name:    pkgsrc
Committed By:   mef
Date:           Thu May  7 00:24:59 UTC 2015

Modified Files:
        pkgsrc/sysutils/coreutils: Makefile Makefile.common distinfo
Removed Files:
        pkgsrc/sysutils/coreutils/patches: patch-lib_parse-datetime.c

Log Message:
(pkgsrc)
  - Drop patches/patch-lib_parse-datetime.c (the comment of the patch was)
      | Fix CVE-2014-9471 TZ parsing bug.
      | Via https://debbugs.gnu.org/cgi/bugreport.cgi?bug=16872
  - Set MAKE_JOBS_SAFE= no, to avoid following problem
      # --- man/chroot.1 ---
      #   GEN      man/chroot.1
      # help2man: can't get `--help' info from chroot.td/chroot
      # Try `--no-discard-stderr' if option outputs to stderr
      # *** [man/chroot.1] Error code 127
(upstream)
  - update to 8.23
------------------
* Noteworthy changes in release 8.23 (2014-07-18) [stable]

** Bug fixes

  chmod -Rc no longer issues erroneous warnings for files with special bits set.
  [bug introduced in coreutils-6.0]

  cp -a, mv, and install --preserve-context, once again set the correct SELinux
  context for existing directories in the destination.  Previously they set
  the context of an existing directory to that of its last copied descendent.
  [bug introduced in coreutils-8.22]

  cp -a, mv, and install --preserve-context, no longer seg fault when running
  with SELinux enabled, when copying from file systems that return an error
  when reading the SELinux context for a file.
  [bug introduced in coreutils-8.22]

  cp -a and mv now preserve xattrs of symlinks copied across file systems.
  [bug introduced with extended attribute preservation feature in coreutils-7.1]

  date could crash or go into an infinite loop when parsing a malformed TZ="".
  [bug introduced with the --date='TZ="" ..' parsing feature in coreutils-5.3.0]

  dd's ASCII and EBCDIC conversions were incompatible with common practice and
  with POSIX, and have been corrected as follows.  First, conv=ascii now
  implies conv=unblock, and conv=ebcdic and conv=ibm now imply conv=block.
  Second, the translation tables for dd conv=ascii and conv=ebcdic have been
  corrected as shown in the following table, where A is the ASCII value, W is
  the old, wrong EBCDIC value, and E is the new, corrected EBCDIC value; all
  values are in octal.

      A   W   E
     041 117 132
     133 112 255
     135 132 275
     136 137 232
     174 152 117
     176 241 137
     313 232 152
     325 255 112
     345 275 241

  [These dd bugs were present in "the beginning".]

  df has more fixes related to the newer dynamic representation of file systems:
  Duplicates are elided for virtual file systems like tmpfs.
  Details for the correct device are output for points mounted multiple times.
  Placeholder values are output for inaccessible file systems, rather than
  than error messages or values for the wrong file system.
  [These bugs were present in "the beginning".]

  df now outputs all appropriate entries in the presence of bind mounts.
  On some systems, entries would have been incorrectly elided due to
  them being considered "dummy" mounts.
  [bug introduced in coreutils-8.22]

  du now silently ignores directory cycles introduced with bind mounts.
  Previously it would issue a warning and exit with a failure status.
  [bug introduced in coreutils-8.1]

  head --bytes=-N and --lines=-N now handles devices more
  consistently, not ignoring data from virtual devices like /dev/zero,
  or on BSD systems data from tty devices.
  [bug introduced in coreutils-5.0.1]

  head --bytes=-N - no longer fails with a bogus diagnostic when stdin's
  seek pointer is not at the beginning.
  [bug introduced with the --bytes=-N feature in coreutils-5.0.1]

  head --lines=-0, when the input does not contain a trailing '\n',
  now copies all input to stdout.  Previously nothing was output in this case.
  [bug introduced with the --lines=-N feature in coreutils-5.0.1]

  id, when invoked with no user name argument, now prints the correct group ID.
  Previously, in the default output format, it would print the default group ID
  in the password database, which may be neither real nor effective.  For e.g.,
  when run set-GID, or when the database changes outside the current session.
  [bug introduced in coreutils-8.1]

  ln -sf now replaces symbolic links whose targets can't exist.  Previously
  it would display an error, requiring --no-dereference to avoid the issue.
  [bug introduced in coreutils-5.3.0]

  ln -sr '' F no longer segfaults.  Now works as expected.
  [bug introduced with the --relative feature in coreutils-8.16]

  numfmt now handles blanks correctly in all unibyte locales.  Previously
  in locales where character 0xA0 is a blank, numfmt would mishandle it.
  [bug introduced when numfmt was added in coreutils-8.21]

  ptx --format long option parsing no longer falls through into the --help case.
  [bug introduced in TEXTUTILS-1_22i]

  ptx now consistently trims whitespace when processing multiple files.
  [This bug was present in "the beginning".]

  seq again generates correct output with start or end values = -0.
  [bug introduced in coreutils-8.20.]

  shuf --repeat no longer dumps core if the input is empty.
  [bug introduced with the --repeat feature in coreutils-8.22]

  sort when using multiple threads now avoids undefined behavior with mutex
  destruction, which could cause deadlocks on some implementations.
  [bug introduced in coreutils-8.6]

  tail -f now uses polling mode for VXFS to cater for its clustered mode.
  [bug introduced with inotify support added in coreutils-7.5]

** New features

  od accepts a new option: --endian=TYPE to handle inputs with different byte
  orders, or to provide consistent output on systems with disparate endianness.

  configure accepts the new option --enable-single-binary to build all the
  selected programs in a single binary called "coreutils".  The selected
  programs can still be called directly using symlinks to "coreutils" or
  shebangs with the option --coreutils-prog= passed to this program.  The
  install behavior is determined by the option --enable-single-binary=symlinks
  or --enable-single-binary=shebangs (the default).  With the symlinks option,
  you can't make a second symlink to any program because that will change the
  name of the called program, which is used by coreutils to determine the
  desired program.  The shebangs option doesn't suffer from this problem, but
  the /proc/$pid/cmdline file might not be updated on all the platforms.  The
  functionality of each program is not affected but this single binary will
  depend on all the required dynamic libraries even to run simple programs.
  If you desire to build some tools outside the single binary file, you can
  pass the option --enable-single-binary-exceptions=PROG_LIST with the comma
  separated list of programs you want to build separately.  This flag
  considerably reduces the overall size of the installed binaries which makes
  it suitable for embedded system.

** Changes in behavior

  chroot with an argument of "/" no longer implicitly changes the current
  directory to "/", allowing changing only user credentials for a command.

  chroot --userspec will now unset supplemental groups associated with root,
  and instead use the supplemental groups of the specified user.

  cut -d$'\n' again outputs lines identified in the --fields list, having
  not done so in v8.21 and v8.22.  Note using this non portable functionality
  will result in the delayed output of lines.

  ls with none of LS_COLORS or COLORTERM environment variables set,
  will now honor an empty or unknown TERM environment variable,
  and not output colors even with --colors=always.

** Improvements

  chroot has better --userspec and --group look-ups, with numeric IDs never
  causing name look-up errors.  Also look-ups are first done outside the chroot,
  in case the look-up within the chroot fails due to library conflicts etc.

  install now allows the combination of the -D and -t options.

  numfmt supports zero padding of numbers using the standard printf
  syntax of a leading zero, for example --format="%010f".
  Also throughput was improved by up to 800% by avoiding redundant processing.

  shred now supports multiple passes on GNU/Linux tape devices by rewinding
  the tape before each pass, avoids redundant writes to empty files,
  uses direct I/O for all passes where possible, and attempts to clear
  inode storage used for small files on some file systems.

  split avoids unnecessary input buffering, immediately writing input to output
  which is significant with --filter or when writing to fifos or stdout etc.

  stat and tail work better with HFS+, HFSX, LogFS and ConfigFS.  stat -f
  --format=%T now reports the file system type, and tail -f now uses inotify,
  rather than the default of issuing a warning and reverting to polling.


To generate a diff of this commit:
cvs rdiff -u -r1.64 -r1.65 pkgsrc/sysutils/coreutils/Makefile
cvs rdiff -u -r1.10 -r1.11 pkgsrc/sysutils/coreutils/Makefile.common
cvs rdiff -u -r1.30 -r1.31 pkgsrc/sysutils/coreutils/distinfo
cvs rdiff -u -r1.1 -r0 \
    pkgsrc/sysutils/coreutils/patches/patch-lib_parse-datetime.c

Please note that diffs are not public domain; they are subject to the
copyright notices on the relevant files.