Обсуждение: BUG #14064: Sort order of bytea, etc. not defined
The following bug has been logged on the website: Bug reference: 14064 Logged by: Chris Pacejo Email address: cpacejo@clearskydata.com PostgreSQL version: 9.5.2 Operating system: any Description: The documentation does not define (nor call out as undefined) the sort order of bytea, bit varying, and other sequence types. While the bytea sort order is unsurprising (a < b if a is a prefix of b, but b > a if a prefix of b > the same length prefix of a), it is not the only such "unsurprising" sort order, and it would be helpful to have the guarantee that this sort order can be relied upon (especially in concert with range types).
Hello, Cpacejo. You wrote: ccc> The following bug has been logged on the website: ccc> Bug reference: 14064 ccc> Logged by: Chris Pacejo ccc> Email address: cpacejo@clearskydata.com ccc> PostgreSQL version: 9.5.2 ccc> Operating system: any ccc> Description: ccc> The documentation does not define (nor call out as undefined) the sort order ccc> of bytea, bit varying, and other sequence types. While the bytea sort order ccc> is unsurprising (a < b if a is a prefix of b, but b > a if a prefix of b > ccc> the same length prefix of a), it is not the only such "unsurprising" sort ccc> order, and it would be helpful to have the guarantee that this sort order ccc> can be relied upon (especially in concert with range types). You shouls use explicit type casting for such types. -- With best wishes, Pavel mailto:pavel@gf.microolap.com
I'm not sure I understand? I know the data is bytea, how would casting it to another type help? (I don't think casting bytea is even possible?) On May 4, 2016 2:16 AM, "Pavel Golub" <pavel@microolap.com> wrote: > Hello, Cpacejo. > > You wrote: > > ccc> The following bug has been logged on the website: > > ccc> Bug reference: 14064 > ccc> Logged by: Chris Pacejo > ccc> Email address: cpacejo@clearskydata.com > ccc> PostgreSQL version: 9.5.2 > ccc> Operating system: any > ccc> Description: > > ccc> The documentation does not define (nor call out as undefined) the > sort order > ccc> of bytea, bit varying, and other sequence types. While the bytea > sort order > ccc> is unsurprising (a < b if a is a prefix of b, but b > a if a prefix > of b > > ccc> the same length prefix of a), it is not the only such "unsurprising" > sort > ccc> order, and it would be helpful to have the guarantee that this sort > order > ccc> can be relied upon (especially in concert with range types). > > > > You shouls use explicit type casting for such types. > > -- > With best wishes, > Pavel mailto:pavel@gf.microolap.com > >