Tests results on Oracle server:

Compared to results posted by Nick, there's more.

Few tests failed because queries syntax are not compatible with Oracle. We start working on those to fix the oracle part:

SQL: select first_names from persons limit !>>>!0, stack trace: 
nsoracle.c:3904:ora_tcl_command: error in `OCIStmtExecute ()':
 ORA-00933: SQL command not properly ended

SQL: select first_names from persons limit !>>>!0
    while executing
"ns_ora 0or1row nsdb0 {select first_names from persons limit 0}"

Type acs_object: select object_id from acs_objects failed:
nsoracle.c:3904:ora_tcl_command: error in `OCIStmtExecute ()':
 ORA-00911: invalid character

SQL: select object_id as !>>>!__pk from acs_objects limit 1

SQL: select a.*, lower(ot.table_name) as obj_type_table 
     from acs_attributes a, acs_object_types ot 
     where ot.object_type = a.object_type 
     order by !>>>!object_type, stack trace: 
nsoracle.c:3904:ora_tcl_command: error in `OCIStmtExecute ()':
 ORA-00918: column ambiguously defined

process_objects_csv (body 0): Error during execution: 
Setup failed with error nsoracle.c:3542:ora_tcl_command: error in `OCIStmtExecute ()':
 ORA-06550: line 4, column 13:
PLS-00306: wrong number or types of arguments in call to 'NEW'
ORA-06550: line 4, column 7:
PL/SQL: Statement ignored

SQL: 
      
    BEGIN
      :1 := person.new(creation_user    => :creation_user,
                                   creation_ip      => :creation_ip,
                                   first_names      => :first_names,
                                   last_name        => :last_name
      );
    END; 

db__caching: we substitute "limit 0" for "where person_id=1 and person_id=2". Results: http://openacs.innova.uned.es/test/admin/testcase?testcase_id=db__caching&package_key=acs-tcl&view_by=testcase&category=&quiet=0

datamodel__acs_object_type_check: we changed the variable name "__pk" to "the_pk". Results: http://openacs.innova.uned.es/test/admin/testcase?testcase_id=datamodel__acs_object_type_check&package_key=acs-tcl&view_by=testcase&category=&quiet=0

datamodel__acs_attribute_check: we changed "order by object_type" to "order by a.object_type". Results: http://openacs.innova.uned.es/test/admin/testcase?testcase_id=datamodel__acs_attribute_check&package_key=acs-tcl&view_by=testcase&category=&quiet=0

Now we will add missing types in the array (added types are between * *):

array set allow_types {
string {TEXT VARCHAR CHAR *VARCHAR2*}
boolean {BOOL INT2 INT4 *CHAR*}
number {NUMERIC INT2 INT4 INT8 FLOAT4 FLOAT8 *NUMBER*}
integer {INT2 INT4 INT8 *NUMBER*}
money {NUMERIC FLOAT4 FLOAT8}
timestamp {TIMESTAMPTZ}
time_of_day {TIMESTAMPTZ}
enumeration {INT2 INT4 INT8}
url {VARCHAR TEXT *VARCHAR2*}
email {VARCHAR TEXT *VARCHAR2*}
text {VARCHAR TEXT CLOB *VARCHAR2*}
keyword {CHAR VARCHAR TEXT *VARCHAR2*}
}

However we don't know what to do with timestamp type. On Oracle timestamp corresponds to the "date" datatype but on postgresql date and timestamp are different datatypes. Any idea will be wellcome.

process_objects_csv: person.new() requires "email" input arg . We added the email field in the csv file. Results: http://openacs.innova.uned.es/test/admin/testcase?testcase_id=process_objects_csv&package_key=acs-tcl&view_by=testcase&category=&quiet=0

If there is no objection, I will commit those fixes.