----- "Thomas Lange" <[email protected]> schrieb:
> >>>>> On Tue, 6 Jul 2010 12:26:13 +0100 (GMT+01:00), > [email protected] said: > > > > As I would think of the setup-storage program, it should exit > with an error on 100% if it is not possible to create such a > partition, because there are already others or what ever. "0-" instead > would run through, even there is not 100% of the disk size available. > > > In my opinion the implementation you use should be considered a > bug since it is not what a user expects and additionally seems to > brake under certain conditions (like mine). > > IMO this is only a minor bug. If you specify 99%- and no other > partition, then setup-storage will put the whole disk size into > this partition. People know, that you cannot use 100% of a disk, > since there's always some overhead. Even the MBR is sort of > overhead. You will also not get 100% inside your file system. > Ok, let's say it this way. As a user, who has partitioning in his mind using tools like fdisk or parted, 100% is not 100% of the disk. It is 100% of the available disk space for a partition. The way it is implemented, it is not the way one would expect since one wants to partition. Imagine the tools console UIs. They do not offer you % normally. But they offer Start and End, where End is predefined by the maximum amount the partition can have (aka. 100% if there is non already). So, to use % in a more complicated scenario: defining part1 = 20%, part2 = 50%, part3 = 30% is not working. Do you think that is what anyone would expect? I can understand that from a bare metal technical view it might be correct --- as you mentioned 100% of the bare disk size can't be used because of overhead. But this is a tool you want to work with and therefore needs to have the user in focus, isn't it? > @Michael: I think it should be ok, to document that people should use > 99%- instead of 100%. I agree. At least you need to document this for the user. But doesn't this proves my point? You need to document it because one would most likely not expect this behavior. Greets > > -- > regards Thomas
