On Mon, Mar 21, 2005 at 12:39:09PM -0800, TJ O'Donnell wrote: > I have N-bit data pairs. I want to write a c-language function > which compares bits set in each. N is typically 512, but could be other. > I can store this as bit varying(512) or bytea(64). I can't decide which. > Here are the questions that concern me. > 1) will each take the same storage?
Bit strings takes 8 bytes + 1 byte for each 8 bits (or fraction thereof) of data; bytea takes 4 bytes + 1 byte for each byte of data. You can find the structure definition for bit strings (VarBit) in <server/utils/varbit.h>; bytea is just a typedef for struct varlena. > 2) can I pass bit varying data to a c-language function? I can't find any > docs or examples of that. See src/backend/utils/adt/varbit.c in the PostgreSQL source code. > 3) are bit strings stored as actual bits or as character strings of 0 and 1? Actual bits, so manipulating the data in C should be the same for both types. Bit strings take an extra 4 bytes of storage, but if that doesn't matter then you might want to consider which type would be easier to manipulate in SQL (bytea functions vs. bit string functions). -- Michael Fuhr http://www.fuhr.org/~mfuhr/ ---------------------------(end of broadcast)--------------------------- TIP 9: the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match
