Parted 1.5.6, stablizing...

[Andrew Clausen]

Hi all,

I've uploaded 1.5.6.  All regression tests pass, all docs are
up to date, and I think we're ready for 1.6.0-pre1.

Yura: I think we should wait before putting reiserfs into a stable
version.  I'll fork off 1.7.x at the same time as 1.6.x, so
hopefully 1.8.x (or whatever) shouldn't be too long coming...
We can always merge back if it's looking really stable. :)

Things that need to happen:
* submit the new parted.pot to get translations.  (They can't do
branches, so that means no more tranlsation updates for 1.4.x,
hence the wait)

* testing outside my bedroom ;)  I should write a HOWTO for getting
the regression tests running

* include the new cool texinfo docs, once they're in sync :)

Change summary:
        updated the user documentation, fixed minor FAT bugs, major MIPS
        partition table bugs, minor changes to UI, fixed all regression tests
        - all pass :)

Change log:
* debug/clearfat/clearfat.c (main): check if fs->type->name is fat16/fat32,
  not FAT
* debug/test/test (ped_test_get_val, ped_test, ped_test_expect_fail):
  now operate in parted's script mode.  Removed support for parsing input
  in stdin (it gets ignored in script mode)
* debug/test/test_*: remove extra parameter for stdin input
* debug/test/test_disk_* (test_outside_boundary): removed... the front end
  deals with this, and obviously works ;)
* debug/test/test_disk_msdos debug/test/test_disk_pc98 (test_primary_overlap):
  use fat16, not fat
* debug/test/test_disk_mips: added
* debug/test/test_fs_ext2 (do_diff, prepare_resize_tests): to make the diff
  "right", make diff ignore lost+found, rather than deleting it before
  the diff.
* debug/test/test_fs_ext2: e2fsck -n... otherwise new version refuse to run
  non-interactively
* debug/test/test_fs_fat (test_mkpartfs): added two tests for creating
  minimum size fat16 and fat32.  Give the constraint solver some work ;)
* libparted/disk_*.c (*_partition_set_system): removed "special" cases
  for partition types, if the special case is the default
* libparted/disk_mips.c (mips_read, mips_write, mips_partition_new,
  mips_partition_get_flag, mips_partition_set_flag):
  MIPSDiskData now uses partition numbers, not pointers to maintain the
  swap/boot/root flags.  There was a dangling pointer bug on removing
  partitions, which was fundamental to the old bogus design.
* (_flush_stale_flags, mips_partition_enumerate): detect dangling partition
  nubers on swap/boot/root flags, and remove them.
* (_handle_no_volume_header, mips_read): deal with missing volume header
  partitions better.
* (_generate_partition): fixed out-by-2 bug... should be ajdusting upwards
  not downwards
* (_get_strict_primary_constraint): renamed from (_get_strict_constraint)
* (_get_strict_primary_constraint): now won't "align" a partition on top
  the volume header
* libparted/disk_pc98.c (pc98_partition_set_flag): removed old cruft wrt
  part->fs_type
* libparted/disk_sun.c (sun_partition_set_flag): removed old cruft wrt
  part->fs_type
* libparted/fs_fat/calc.c libparted/fs_fat/calc.h
  (fat_min_cluster_size, fat_max_cluster_size): added
* (calc_sizes, fat_calc_sizes): replaced first parameter (geom) with a length
* (calc_sizes): make the "fat_space" adjustment only if the old value was
  impossible.  (This is what Windows does FWICT, but I need to collect more
  info)
* (fat_calc_sizes, fat_calc_resize_sizes): iterate between
  (fat_min_cluster_size, fat_max_cluster_size) 
* libparted/fs_fat/fat.c (fat_create): call (fat_table_set_cluster_count),
  so the free cluster count is calculated properly on FAT32
* (_test_resize_size): renamed from (_test_size)
* (_get_min_resize_size): renamed from (_get_min_size)
* (fat_get_create_constraint_fat16, fat_get_create_constraint_fat32):
* (_get_min_resize_size): renamed from (_get_min_size)
* (fat_get_create_constraint_fat16, fat_get_create_constraint_fat32):
  plugged in magic constants instead of computing.  Computing is hard, and
  the hack used is commented out (it's fragile).  Should be fixed...
* libparted/fs_fat/resize.c (fat_resize, free_root_dir): when the fs
  is converted fat32 -> fat16, the root directory was been duplicated,
  but not referenced (fat16 has a separate area for the root dir).  This
  lead to these clusters appearing as "lost clusters" in dosfsck, etc.
  Now, we free the root directory.
* parted/parted.c (_grow_over_small_freespace): only grow over freespace
  for big partitions
* parted/parted.c (_done): give warnings about lilo and /etc/fstab
* (do_mkpart): made the file system parameter optional
* doc/USER: updated to match 1.5.x

Andrew