* Added dialog (accessable from 'gparted' menu) which shows supported operations for the various filesystems.
I plan to put some more info in it, like what is missing etc..
* Refined checks for functionality a bit. e.g. even if a filesystem can't be read, it's still possible to grow it
(but shrinking gets disabled). Also removed that stupid grow_only flag from Utils.h and reordered Win_GParted::Set_Valid_Operations()
* added support for jfs (create, copy, grow) and hfs (create, copy)
* fixed a bug in copying xfs filesystems.
* fixed a number of small bugs/annoying issues.
* Added support for xfs. this means creating and growing xfs filesystems. Shrinking requires some hacking with dump_xfs etc..
i'll add that at a later point. :)
* Replaced boolean 'resize' with 'shrink' and 'grow'. It seems some filesystems only support growing (e.g. xfs) so i need
two flags to control the behaviour of the filesystemclasses.
* include/GParted_Core.h,
src/GParted_Core.cc: modified layoutscan a bit for improved handling of metadata.
* include/Operation.h,
src/Operation.cc: modified Insert_Unallocated() to prevent potential crasher.
* src/ext2.cc,
src/ext3.cc,
src/ntfs.cc,
src/reiserfs.cc: added LANG=C to the commands which read the used space of a filesystem. This prevent errors with some locales.
* added MIN and MAX to filesystemstruct to set min. and max sizes of a filesystem. So instead of checking per filesystem
i now simply check the fs.MIN or fs.MAX. this results in less and cleaner code. Also this will come in handy when adding
support for new filesystems. (This also fixed several minor bugs with filesystemsizes and gained some improvement in resizer
performance)
* src/GParted_Core.cc: When ( ! deep_scan ) now only one device is scanned (first found one). Shaves off a couple of millisec from startuptime |-)
* src/Win_GParted.cc: Fixed small bug with removable devices and the devicemenu. Also fixed redundant calling of 2 functions.
* include/FileSystem.h,
src/FileSystem.cc: Execute_Command() now returns exit status of executed command
* the filesystemclasses: returnvalue (bool) is now set according to the return status of the command
* src/GParted_Core.cc: It seemed some kernels remove and re-add blockdevicepaths after the commit_to_os. This caused trouble at some points
where the path was required. I decided to sleep 1 sec after each scan. So rescanning devices will take 1 extra second per device :|
* src/Win_GParted.cc: fixed typo in string.
* Every devicescan now tests on beforehand if the kernel is able to reread the partitiontable. If the kernel is unable to do
this, i disallow most operations. This may seem weird and even a bad thing to do, but the fact is it protects
the innocent user from a lot of potential damage. Till the linuxkernel is able to reread partitiontables no matter what, this seems
to be the best option. Of course a dialog with information will popup whenever such a situation is encountered.
In a next release i might consider adding an 'advanced mode' for users who know what they're doing.
Bart Hakvoort