[Thunar-dev] Minor things part $PREVIOUS++

[Benedikt Meurer]

Harold Aling wrote:
>>Ah, you mean, Thunar should first check whether enough space is
>>available? Well, even if you ignore the numerous race conditions that
>>would arise here, you'd still be unable to reliably detect whether or
>>not enough space is available (i.e. different block sizes, available
>>inodes, user/group quotas, etc.). Thunar would probably reject to copy
>>files based on its estimation, while it would in reality be possible to
>>copy the files. Even if it would be possible to take care of all
>>conditions on all supported platforms, it'd require atleast a year or
>>two to come up with a working implementation. Really not something I'd
>>want to waste my time with.
> 
> Oops, sorry... Since Thunar is able to report free space on a mount, i
> thought it shouldn't be that difficult to implement a 'cancel/continue
> anyway' dialog...

Unix file systems are complex. The free space reported by Thunar should
be considered as a rough measurement. A simple example: You can easily
fill up 4MB with thousand 1B files on a file system with a block size of
4KB (not taking into account the space required for the thousand
inodes). This depends heavily on the operating system and the file
system (i.e. some modern file systems will probably place the 1B data
into the inode).

Benedikt