SAP Note 1178671-Conversion of Views from non-Unicode to Unicode

Note Language: English     Version: 7     Validity: Valid Since 2008.06.13

Symptom
When converting a system from non-Unicode to Unicode one of the following
issues can occur with customer defined views:
  o A) the import of a package fails with messages:
(DB) ERROR: DDL statement failed
(CREATE VIEW "SOME_VIEW_NAME" ....
typically package SAPVIEW is affected which contains all view
definitions.
  o B) after conversion, in the Unicode system, views do not behave as
expected, and are inconsistent with their pre-conversion behaviour.
Other terms
SQL DDL view SAPVIEW.STR Unicode conversion R3load
Reason and Prerequisites
During a Unicode conversion R3ldctl reads the source system and generates
an .STR file, typically SAPVIEW.STR, which contains a specification of each
view, consisting of sections:
view: <name of view>
fld: <several lines, each specifying a field>
qry: SELECT ... FROM ... WHERE ...
The WHERE part contains an expression with text constants, which contain
non-7bit-ASCII charcters, eg
... AND T0001."CITY" = 'MÜNCHEN'
At import time R3load reads this .STR file, and creates the views in the
destination system; as this is a Unicode R3load it expects the .STR file to
be UTF-8 encoded. This can lead to
  o the generation of syntactically incorrect DDL statements, resulting
in symptom A);
  o successful creation of a view, but with select semantics different
from the source view, which is symptom B).

Solution
Depending on the symptom:
    o A) import error
- modify the offending .STR file, by removing the problematic
character or replacing it, in the WHERE section indicated by
the error message.
- redo the import of the package, which should now succeed.
- take a note of the view name, for post-conversion processing.
    o B) unexpected view selectivity
- identify the view, by unexpected behaviour you observe.

- or you may apply the attached awk script "8bitscript" to the
.STR file in question; it lists all lines containing text with
"8th bit on" characters. Typically these will occur in the
WHERE sections of view definitions which may be corrupted by
the import.
- instead of the Unix platform tool awk, you may use the platform
independent SAP command sapiconv (cf Note 752859) to find
offending non-7bit-ASCII characters in the .STR file. A
possible call is, where 1101 is the 7bit-ASCII code page; in
the output file you will find the .STR input up to the first
offending character:
.../sapiconv -f 1101 -t 1100 < ..../SAPVIEW.STR >yourTempFile
- process any questionable views as described in section "Post
Processing" below.
POST PROCESSING: recreate the view from the Database version:
    o transaction SE11: select the view, and check it; this should show
an inconsistency of DD- and DB-version
    o transaction SE14: delete the database view
    o transaction SE11: activate the view.

Final Solution
Generally it is impossible and/or incompatible with existing system copy
tool interfaces to generate view descriptions with a defined non-Unicode
encoding. Hence we will support this scenario by:
    o a conversion preparation tool which finds views with problematic
WHERE clauses.
    o a conversion post-processing tool which finds views with Database/
Datadictionary inconsistencies, and recreates the Database view.
These improvements will become available with Support Packages, the
identification of which will be listed here upon availability.

-------------------------------------------------------------------------------------------------------------------

Release Status: Released for Customer
Released on: 2008.06.13 08:14:39
Master Language: English
Priority: Recommendations/additional info
Category: Upgrade information
Primary Component: BC-I18-UNI I18N Unicode
Secondary Components:
BC-INS-MIG DB/OS Migrations

 

The Note is release-independent

Related Notes

Number       Short Text
1319517     Unicode Collection Note

File Type     File Name        Language        Size
TXT         8bitscript.txt          E                1 KB