[sqlalchemy] Order of constraints at creation time
[This is a repost, direct mail wasn't working] Hi all, I spent a moment at fixing here and there to make some test work under firebird. I'm out of luck figuring out the proper solution for engine.ReflectionTest.testbasic: FB2 fails here because the test tries to create a table with a primary key and a foreign key to itself, but the issued SQL define the foreign key sooner than that the primary key. In other words, the statement is something like CREATE TABLE engine_users ( user_id INTEGER, ... FOREIGN KEY (parent_user_id) REFERENCES engine_users (user_id) PRIMARY KEY (user_id) ) This works for example in sqlite, but AFAICS FB2 is not smart enough: it accepts the stmt only if I swap the constraints, otherwise it complains about a missing unique index on the target field... Digging the issue, I was looking at AnsiSchemaGenerator.visit_table() which, at some point, calls .get_column_specification() passing a first_pk flag, which at first seemed what I was looking for: as the comment suggests, if at all possible the primary key should go earlier on the column itself, but: a) none of the current backends make any use of it b) even if they did, how could they omit the primary key constraint from the final loop on the table's ones? Ideally the PK constraint (if any) should be visited before the remaining FK constraints, but I don't see any sensible way of doing that, since the container is a set. Any advice? Thanks in advance, and happy gnu year everybody ;) ciao, lele. --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups sqlalchemy group. To post to this group, send email to sqlalchemy@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sqlalchemy?hl=en -~--~~~~--~~--~--~---
[sqlalchemy] Order of constraints at creation time
Hi all, I spent a moment at fixing here and there to make some test work under firebird. I'm out of luck figuring out the proper solution for engine.ReflectionTest.testbasic: FB2 fails here because the test tries to create a table with a primary key and a foreign key to itself, but the issued SQL define the foreign key sooner than that the primary key. In other words, the statement is something like CREATE TABLE engine_users ( user_id INTEGER, ... FOREIGN KEY (parent_user_id) REFERENCES engine_users (user_id) PRIMARY KEY (user_id) ) This works for example in sqlite, but AFAICS FB2 is not smart enough: it accepts the stmt only if I swap the constraints, otherwise it complains about a missing unique index on the target field... Digging the issue, I was looking at AnsiSchemaGenerator.visit_table() which, at some point, calls .get_column_specification() passing a first_pk flag, which at first seemed what I was looking for: as the comment suggests, if at all possible the primary key should go earlier on the column itself, but: a) none of the current backends make any use of it b) even if they did, how could they omit the primary key constraint from the final loop on the table's ones? Ideally the PK constraint (if any) should be visited before the remaining FK constraints, but I don't see any sensible way of doing that, since the container is a set. Any advice? Thanks in advance, and happy gnu year everybody ;) ciao, lele. --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups sqlalchemy group. To post to this group, send email to sqlalchemy@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sqlalchemy?hl=en -~--~~~~--~~--~--~---
[sqlalchemy] Order of constraints at creation time
Hi all, I spent a moment at fixing here and there to make some test work under firebird. I'm out of luck figuring out the proper solution for engine.ReflectionTest.testbasic: FB2 fails here because the test tries to create a table with a primary key and a foreign key to itself, but the issued SQL define the foreign key sooner than that the primary key. In other words, the statement is something like CREATE TABLE engine_users ( user_id INTEGER, ... FOREIGN KEY (parent_user_id) REFERENCES engine_users (user_id) PRIMARY KEY (user_id) ) This works for example in sqlite, but AFAICS FB2 is not smart enough: it accepts the stmt only if I swap the constraints, otherwise it complains about a missing unique index on the target field... Digging the issue, I was looking at AnsiSchemaGenerator.visit_table() which, at some point, calls .get_column_specification() passing a first_pk flag, which at first seemed what I was looking for: as the comment suggests, if at all possible the primary key should go earlier on the column itself, but: a) none of the current backends make any use of it b) even if they did, how could they omit the primary key constraint from the final loop on the table's ones? Ideally the PK constraint (if any) should be visited before the remaining FK constraints, but I don't see any sensible way of doing that, since the container is a set. Any advice? Thanks in advance, and happy gnu year everybody ;) ciao, lele. --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups sqlalchemy group. To post to this group, send email to sqlalchemy@googlegroups.com To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/sqlalchemy?hl=en -~--~~~~--~~--~--~---
