Kbase 214: CRC: Impact of deleting index
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
10/05/1998 |
|
CRC: Impact of deleting index
920217-elp04
Subject: CRC Option: Problems when index is deleted.
In the process of deploying a database schema, a user may have
deleted an index in the db, and then dumped the schema and loaded
it into an empty db. The CRC's are considerably different on
the two now "identical" db's. The reason for this is that by
deleting an index, there was a shift in index-num. This is used
in the CRC calculation. It is also used in the .r code, and that
is why it is used in the calculation.
When the definitions are dumped, they aren't necessarily written to
the .df file in the order in which the dictionary information was
originally added, since the files are listed alphabetically by
filename, so we could seem some CRC differences there as well.
Users that are planning on using CRC should follow these steps.
1) Create the basic db.
2) Dump out the data definitions from the development db.
3) Reload the data definitions into the basic db.
(so that it will get loaded in the same order as the customers).
4) Compile their application against the newly loaded db.
5) Create the customer's db using the same .df as used in step 3.
6) Maintain the basic db and the customer's db using incremental
definitions (delta.df files)
This way the customer's CRC's will match the VAR's.
The above applies to V6 databases. In Version 7, the metaschema
field _Field._field-rpos may have a similar impact during deployment.
So the steps outlined above should be followed in both versions.
Progress Software Technical Support Note # 214