[parted-devel] [PATCH] Properly sync partitions with operating system

[Jim Meyering]

Petr Uzel <petr.uzel at suse.cz> writes:

> On Fri, Feb 20, 2009 at 01:01:03PM +0100, Joel Granados wrote:
>> > merely queries the code for the appropriate partition table type
>> > and always returns a non-negative number.
>>
>> well.  The specific code in each label might return a negative number
>> for error as well.   Think of some spec that defines that maximum number
>> as something that might be read from disk (just playing the devils
>> advocate here).  The read from disk might fail and then it must be able
>> to tell the calling function that something when wrong.
>>
>> >
>> > > when it encounters an error.  All possitive numbers can be valide to
>> > > express max_number_of_supported_partitions.  Including 0. So a negative
>> > > number would seem natural for me to use in this case.
>> > >
>> > >  0: means that the label does no support partitions.
>> > >  <0: means the number of partitions
>> > >  >0: means something nasty has ocurred.
>
> AFAIK kernel has similar function that returns 1 if the device can not
> be partitioned (maximum of one partition on a device does not make
> sense anyway). So 0 could be used to indicate an error.
>
> But this would be incosistent with the rest of functions that return
> -1 in case of error and thus it might make things worse.

The other functions are either wrong to return "int"
(if they always return non-negative) or have to handle
failure indicated by returning a negative value.

This is neither of those cases.  It's essentially returning
a per-partition-type non-negative constant.
So it can safely return "unsigned int".