Informix

nested-group-icon.png

DB2

Best practice to insert CLOB/BLOB with generic ESQL/C code

  • 1.  Best practice to insert CLOB/BLOB with generic ESQL/C code

    Posted Thu June 04, 2020 09:11 AM
    Hello,

    What is the best solution to insert CLOB/BLOB values from ESQL/C code, when you don't know the storage options of CLOB/BLOB of the SQL columns, because you write generic code that can execute any kind of SQL statement?

    To insert a CLOB/BLOB, you create / initialize an ifx_lo_t object, by using ifx_lo_def_create_spec() / ifx_lo_create()...

    So far I do the following (shorten version edited by hand, without error checking):
    ifx_lo_create_spec_t *cs;
    ifx_lo_t *lo = (ifx_lo_t *) v->sqldata; // mallocated ...
    int lofd;
    ifx_lo_def_create_spec(&cs);
    lofd = ifx_lo_create(cs, LO_RDWR | LO_LOCKALL, lo, &err);
    ifx_lo_write(lofd, buf, len, &err);
    ifx_lo_close(lofd);
    ifx_lo_spec_free(cs);

    According to the doc, this should take "storage characteristics from the inheritance hierarchy" ...

    Is this ok?

    Is it better to use the "system-specified storage characteristics", passing NULL for the creation specs?

    What about the LO_* flags like LO_LOCKALL ...?

    Should I make all these creation options configurable?

    Thanks!
    Seb

    ------------------------------
    Sebastien FLAESCH
    ------------------------------