x/deb-oslo.db
Code Issues Proposed changes

Changed processing unique constraint name.

Browse Source 
Function `_raise_if_duplicate_entry_error()` modified respectively to
new unique constraint name convention.

We found that current constraint name convention allows us create
constraints with duplicate names. It happens if we will add constraints
in different tables with same column names (for example, `name`,
`deleted`). In this case we can not create new constraint due to mysql
limitation (it requires unique constraint name for database)
To solve this issue unique constraint name convention was changed
from "uniq_c1_x_c2_x_c3" to "uniq_t$c1$c2$c3" where `t` it is table name
and columns `c1`, `c2`, `c3` are in UniqueConstraint.

Function `_raise_if_duplicate_entry_error()` parse error from
database and provides us correct column names in error message.
So now we receive correct error message from function.

Change-Id: I4a507deba97c499fcd9738d980e36ca0f9454b71
Fixes: bug 1182054
This commit is contained in:
Victor Sergeyev
2013-05-21 11:16:42 +03:00
parent 69a54905cc
commit 648af43b91
1 changed files with 4 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
openstack/common/db/sqlalchemy/session.py
View File

@@ -412,14 +412,15 @@ def _raise_if_duplicate_entry_error(integrity_error, engine_name):
"""
def get_columns_from_uniq_cons_or_name(columns):
# note(boris-42): UniqueConstraint name convention: "uniq_c1_x_c2_x_c3"
# means that columns c1, c2, c3 are in UniqueConstraint.
# note(vsergeyev): UniqueConstraint name convention: "uniq_t$c1$c2"
# where `t` it is table name and columns `c1`, `c2`
# are in UniqueConstraint.
uniqbase = "uniq_"
if not columns.startswith(uniqbase):
if engine_name == "postgresql":
return [columns[columns.index("_") + 1:columns.rindex("_")]]
return [columns]
return columns[len(uniqbase):].split("_x_")
return columns[len(uniqbase):].split("$")[1:]
if engine_name not in ["mysql", "sqlite", "postgresql"]:
return