"Drew, Stephen" <[EMAIL PROTECTED]> wrote: > Hello all, > > I have written a function to replicate the Oracle DECODE function, which > I register whenever I open a SQLite3 (3.2.6) connection (see bottom of > email). > > However it seems that certain decodes fail due to corrupted data. I > have stepped through my code and seen that the value is correctly set > when the decode function is called, but when the value is retrieved > using sqlite3_column_value, it comes back as gibberish. > > This does not always happen - the particular case involves five decode > statements in the SELECT list. If I move some of the DECODE statements > around or remove them, it works ok. > > So - here are some questions: > > * - Is it safe to call sqlite3_set_result_string( ) as I am doing, i.e. > with the value returned from sqlite3_value_text on the arguments? (I > have included my decode function at the bottom of this email, and > happily release into the public domain....).
There is no such API as sqlite3_set_result_string(). There was an API sqlite_set_result_string() in SQLite version 2, which you appear to be using. But it is certainly not safe to mix APIs. I'm surprised this does not segfault on you. Use sqlite3_result_text() instead. See http://www.sqlite.org/capi3ref.html for additional information. And/or use the function implementations in the func.c source file http://www.sqlite.org/cvstrac/fileview?f=sqlite/src/func.c as examples. -- D. Richard Hipp <[EMAIL PROTECTED]>
