Description:
------------
After migrating to UTF8, database provides two-way semantics for text data types: byte (for single byte encodings is a synonim of char) and char.
https://docs.oracle.com/cd/B28359_01/server.111/b28318/datatype.htm#CNCPT012

Length field in oci_bind_by_name() function forcely expect byte data size.
For now we should update documentation noting it.
For future development, ma be it is reasonable to add a modifier to switch between char and byte semantics.

SQL> CREATE TABLE EXAMPLE (
	MESSAGE		VARCHAR2(16 CHAR) NOT NULL
);

Oracle's client locale is:
NLS_LANG=RUSSIAN_RUSSIA.AL32UTF8
Test script:
---------------
<?php
  require_once("../inc/conn.inc");
  $conn = conn("UPO");
  $var = "тест-тест";

  $stid = oci_parse($conn, 'INSERT INTO EXAMPLE VALUES (:bv)');
  oci_bind_by_name($stid, ':bv', $var, 9);
  if (oci_execute($stid)) {
    echo ("insert OK");
    oci_free_statement($stid);
    oci_close($conn);
  }
  else
    echo ("insert failed");

?>
Expected result:
----------------
If client's locale is NLS_LANG=RUSSIAN_CIS.CL8ISO8859P5 properly encoded (translated from UTF8 back to ISO-8859-5) script works fine.

I expect the same for modern UTF8 locale.

Or at least proper description of issued function: what and how should be corrected.
Actual result:
--------------
If client is configured to use UTF8 locale, quoted script failes with ORA-01461 (can bind a LONG value only for insert into a LONG column) error.
Originally it was find for binding output variables for call of PL/SQL procedure.
In that case the error message was ORA-06525: Length Mismatch for CHAR or RAW data