On Mar 9, 2010, at 1:42 AM, Martin Evans wrote:
There is however one possible area of ambiguity with respect to
bytes or
characters (within a DBD). DBD::ODBC counts the $length as characters
not bytes as how it is eventually encoded in Perl should not matter.
So
if you specify $length=10 and the output parameter is a wide (MS's
term
in ODBC for a unicode character that occupies 2 bytes) character type
DBD::ODBC allocates twice as much plus 2 (for the terminating nul chr)
bytes then encodes it in UTF-8 for Perl - so you'd get at most 10
chrs.
In this case if the output parameter was 11 chrs in length you'd get a
data right truncation error again.
I'm not sure what DBD::Oracle does wrt to chrs or bytes.
This clarifies it a lot for me. The data_length of the column in
Oracle (it is a default number() size) is 22 bytes, so that's what I
have to set it in my code.
--
Bruce Johnson
University of Arizona
College of Pharmacy
Information Technology Group
Institutions do not have opinions, merely customs
