fix for MySql INSERTS with autoincrement #32
Conversation
|
+1 |
adamdubiel
left a comment
adamdubiel
left a comment
There was a problem hiding this comment.
I think changes that you introduced change public APIs too much. From what i understand problem is that you shouldn't set 0 in autoincremented fields? Are all those changes necessary to achieve this?
| */ | ||
| public boolean isSequenceValueSet() { | ||
| return sequenceValueSet; | ||
| void setArgument(String fieldName, Object value) { |
There was a problem hiding this comment.
InsertQuery was supposed to have fluent API, adding this delegator breaks everything - why can't it be done gracefully?
There was a problem hiding this comment.
Chyba nie rozumiem, nie dodawałem żadnych delegatorów. Co dokładnie 'breaks everything'?
Jedyna zmiana w API to usunięcie możliwości ustawiania ręcznie ID, więc 2 metod: sequenceValue(long value) i isSequenceValueSet(), ponieważ nie byłem w stanie ich zachować po wydzieleniu klasy InsertWithAutoincrement.
Te metody są a) niepotrzebne, b) nie mają sensu dla MySql, btw kto chciałby robić coś takiego ?
InsertQuery query = javersPolyJDBC.query().insert().into("my_table")
.value("my_field", 1)
.sequence("my_id", "ANYTHING !")
.sequenceValue(5);
Jeśli nie chcę używać sekwencji to po prostu ich nie używam i ustawiam ID wprost:
InsertQuery query = javersPolyJDBC.query().insert().into("my_table")
.value("my_field", 1)
.value("my_id", "5");
| @Override | ||
| public long generateKey(String sequenceName, Transaction transaction) throws SQLException { | ||
| return DEFAULT_VALUE; | ||
| throw new RuntimeException("not implemented"); |
There was a problem hiding this comment.
could you expand this message a bit?
There was a problem hiding this comment.
will change to throw new RuntimeException("Not implemented. Can't generate key on AutoIncremented");
| /** | ||
| * Manually set sequenced field value. | ||
| */ | ||
| public InsertQuery sequenceValue(long value) { |
There was a problem hiding this comment.
you removed public method - this is a breaking change, is it neccessary?
There was a problem hiding this comment.
I don't think this method is used by polyjdbc clients.
Typically you run queries via polyjdbc QueryRunner:
InsertQuery query = javersPolyJDBC.query().insert().into(getSnapshotTableNameWithSchema())
.value(SNAPSHOT_TYPE, cdoSnapshot.getType().toString())
.value(SNAPSHOT_GLOBAL_ID_FK, globalIdPk)
.value(SNAPSHOT_COMMIT_FK, commitIdPk)
.value(SNAPSHOT_VERSION, cdoSnapshot.getVersion())
.value(SNAPSHOT_STATE, jsonConverter.toJson(cdoSnapshot.getState()))
.value(SNAPSHOT_CHANGED, jsonConverter.toJson(cdoSnapshot.getChanged()))
.value(SNAPSHOT_MANAGED_TYPE, cdoSnapshot.getManagedType().getName())
.sequence(SNAPSHOT_PK, getSnapshotTablePkSeqWithSchema());
return javersPolyJDBC.queryRunner().insert(query);
and QueryRunner increments the sequence and saves the last value. Why to bypass the flow?
| * Insert next sequence value into column of given name. Only one sequenced | ||
| * column per table is supported so far. | ||
| */ | ||
| public InsertQuery sequence(String sequenceField, String sequenceName) { |
There was a problem hiding this comment.
after these changes API will look quite ugly:
InsertQuery query = QueryFactory.instert();
((InsertWithAutoIncrement) query()).sequence(.....)
There was a problem hiding this comment.
no, this API is not changed, the sequence() method is still in the same place :)
No description provided.