* Linux has a MAJOR quirk in that labels such as "Contains Space"
get converted to "/dev/disk/by-label/Contains\x20Space".
* While Rufus already had smart label Syslinux handling (by replacing
the ones found in isolinux.cfg to their more limited FAT32 version)
there was absolutely NO WAY of guessing this quirk until ISOs
such as Red Hat 7 or CentOS 7 started to use spaces in their ISO
labels and people reported breakage!
* This commit also fixes an issue where psz_fullpath was modified when
it shouldn't, which broke Debian 7.7.0, and most likely any ISO
containing a subdirectory in the directory where a syslinux/isolinux
config file is located.
* Closes#396
* Closes#394
* Closes#388
* Closes#380
* Fix label control being reduced in size when a translation modifies it
* Minute adjustments to make the application look better on Windows 8
* Closes#356
* Fix missing DD-Image option when no drive is present
* Prevent formatting of a drive when using an image residing on same drive
* Fix ignored bootable checkbox when using a DD-Image
* Fix wrong title for error messages (due to too many concurrent loc messages)
* Add logic to mount/remount a DD image if it contains a partition we can handle
* Clear GPT structures before copying DD image in case it doesn't overwrites them
* Add a call to refresh the drive layout after copying a DD image
* Move the unlocking of logical drive further down. This may help with the
infamous "device in use by another process" error.
* With that obscene an amount of pre-releases of 6.03 (seriously, how
many more YEARS pre-release of the same version do you actually need?),
of course the syslinux people have managed to break the last remnants
of compatibility they had between a single major version, so we are now
forced to provide a smorgasbord of pre-release and out of band syslinux
binaries on our server, and make sure we detect and handle incompatible
syslinux versions clientside...
For instance, even after fixing the EFI vs isolinux issue, we find
that tails 1.1 is incompatible with 'pre19'. But it also uses its own
'20131220' extended identifier, instead of 'pre1', its closest
relative. So we have to multiply files and symbolic links to try to
keep everybody happy.
Talk about a MAJOR LETDOWN from the syslinux project...
* Closes#363
* It looks like some disks may report a sector size of zero
* Hopefully, this will address #352 (but some report about the
detected sector size from people observing the crash would be nice)
* The call for the node connection information will fail for device such as Mp3 media players
* In that case, we can still obtain the VID:PID from the Device ID
* Also improve mutex release for commandline hogger
