ERP 2.50:Developers Guide/Common Issues, Tips and Tricks
Common Issues, Tips and Tricks
This part of the developers guide is dedicated to discussing common issues and answering frequently asked questions. It tries to provide tips and tricks for common obstacles encountered when developing and extending Openbravo.
Java Development Issues, Build Path problems
To check for a cause and solution check the Setup for Eclipse guide and specifically the section on Build Path problems.
Data Access Layer
Getting warning: Unbalanced calls to enableAsAdminContext and resetAsAdminContext
This warning is displayed (together with a stacktrace) when setting administrator mode is not done correctly. See this part of the developers guide.
How to call a Stored Procedure/Process through the DAL
See these notes in the howto on calling a process through the DAL.
How to read data visible only from a specific Organization
The default DAL behavior (of OBQuery and OBCriteria) is to read information from all readable organizations. This is all organizations to which the current role has access. However, sometimes you want to read information which is only accessible/visible from a specific organization. The code snippet below shows how this can be accomplished:
final OBCriteria<PaymentTerm> obc = OBDal.getInstance().createCriteria(PaymentTerm.class); obc.add(Expression.in("organization.id", OBContext.getOBContext().getOrganizationStructureProvider().getNaturalTree(orgId))); obc.setFilterOnReadableOrganization(false); final List<PaymentTerm> pts = obc.list();
This code snippet shows that you have to add an in expression to the OBCriteria and then disable the standard organization filtering. A similar approach can be implemented for OBQuery.
When flushing/committing a trigger throws an exception, but I can't see/display the real trigger message
Hibernate and JDBC will wrap the exception thrown by the trigger in another exception (the java.sql.BatchUpdateException) and this exception is sometimes wrapped again. Also the java.sql.BatchUpdateException stores the underlying trigger exception in the nextException and not in the cause property. The following method can help you to get to the underlying trigger message:
private String getExceptionMessage(Throwable t) { if (t.getCause() instanceof BatchUpdateException && ((BatchUpdateException) t.getCause()).getNextException() != null) { final BatchUpdateException bue = (BatchUpdateException) t.getCause(); return bue.getNextException().getMessage(); } return t.getMessage(); }
Test your HQL: the HQL Query Tool
Openbravo forge contains a module which offers a query tool which allows you to try a HQL query directly in the Openbravo web interface. The HQL Query Tool is also available through the central repository. See here for the manual of this utility module.
How can I see what SQL is executed?
To see the SQL queries fired by Hibernate, the following log4j property has to be set (in the WEB-INF/classes/log4j.properties or better in openbravo/src/log4j.properties):
### log just the SQL log4j.logger.org.hibernate.SQL=debug
If it does not work for you, it is possible that the log4j.properties file is read from another location than you expect. To see where log4j reads its properties from set this jvm parameter:
-Dlog4j.debug
My query does not return my newly created object!
This can occur when you do not explicitly set the active property to true in the new business object.
The main Openbravo query classes (OBCriteria and OBQuery) automatically add different filter criteria. One of the filter criteria is the active property. As a default the query classes will only return objects with active is true (or 'Y' in the database).
When creating a business object the active property has to be explicitly set to true (setActive(true)). This ensures that the active-filter will not miss it in queries.
My query does not return anything!
There are cases where Hibernate will perform an inner-join when executing a query. If there is no related data in the joined table then the complete query will return nothing.
As an example take the following query:
SELECT p.id, p.image FROM Product AS p
This query seems pretty simple. However, in fact the image is a seperately persisted object and a many-to-one reference from the product to the image table. This query therefore results in an inner-join to the image table. In this case that table was empty and nothing was returned.
To solve this issue, the image has to be joined in an outer join:
SELECT p.id, p.image FROM Product AS p LEFT OUTER JOIN p.image
not-null property references a null or transient value: *.creationDate
This exception is thrown when you try to save a new business object with its Id set. It makes sense to set an id if you want to control the id-setting (and not let hibernate create one for you). However, if the id is set then Hibernate/Openbravo will think it is an existing object and not set the audit info properties.
To solve this also call setNewOBObject(true) on the new business object. This signals to Openbravo that the object is new eventhough its id is set.
not-null property references a null or transient value: *.updated
Same issue as the previous one.
org.hibernate.TransientObjectException: object references an unsaved transient instance - save the transient instance before flushing: ....
This exception occurs when you save an object which refers to another object which has not yet been saved. It applies to so-called many-to-one references, for example from Order to Currency (many orders use the same currency). Openbravo maps the many-to-one associations without cascading. This is for performance reasons.
To prevent this exception save the refered-to instance before the refering instance.
Note that in practice this exception does not happen that often as most references are from transactional to master data, and master data is traditionally setup before transactional data is created.
The import data/REST webservices take care of persisting objects in the correct order so therefore this exception does not occur when importing data through these processes.
Performance: getting the id of a BaseOBObject
To improve performance of single-ended associations, the DAL makes use of the Hibernate proxy functionality. The Hibernate proxy functionality wraps an object inside a Hibernate (cglib) proxy object. The Hibernate proxy object takes care of loading the business object when it is actually accessed and not before.
In many cases a developer just wants access to the id or entityname of an object. To prevent loading of the business object when retrieving just this information, the DalUtil class in org.openbravo.dal.core offers a getId method and a getEntityName method. These methods work directly with the HibernateProxy object and do not load the underlying business object.
Classloading
The DAL loads classes when initializing the DAL. The DAL as a default uses the context class loader of the thread. In some cases this does not work correctly (for example when using the DAL in Ant). The DAL uses the OBClassLoader class to make the classloader configurable. By calling OBClassLoader.setInstance with your own OBClassLoader you can control the class loader used by the DAL.
NPE when calling a getter of a Business Object
All getters of Openbravo business objects return an Object. This is also the case for primitive type properties. In this way Openbravo can support null values in the database (will result in a null value for the property). However, null values result in NPE's when used directly in expressions. For example take this code:
SystemInformation sys = OBDal.getInstance().get(SystemInformation.class, "0"); return sys.isEnableHeartbeat();
When the value of 'enableHeartbeat' is null in the database then the call to isEnableHeartbeat will result in a NPE eventhough the sys variable is set correctly.
The same will happen when using results of getter-calls directly in numeric expressions.
DAL Queries do not return or do not see changes in the database
This can occur in the following situation:
- your program reads objects from the database using the DAL
- then you update the database directly without going through the DAL (either through direct sql or through a stored procedure), the database updates the objects/data you read through the DAL in step 1
- then your program uses the same objects as read in step 1 or you read them again from the database
The problem you encounter then is that the objects do not reflect the database state (they are not up-to-date). This is caused by the following:
When Hibernate reads data from the database it is stored in a cache for that session. This session cache is not auto-updated when data changes in the database outside of your hibernate session. So when in the last step you read the objects again from the database, you get the objects from the session-cache and not from the database.
There are 2 solutions:
- clear the session cache completely
OBDal.getInstance().getSession().flush(); OBDal.getInstance().getSession().clear();
- force reread of specific objects
OBDal.getInstance().getSession().refresh(myObject);
The first approach makes sense if there are many updates in the database or you don't know exactly which objects have changed. The second approach makes sense if you know exactly which object(s) have changed and there are not that many.
In case of the first solution you have to note that objects that have been read (through the DAL) before flushing/clearing can not be used anymore after flush/clear. These objects need to be re-attached to the session or read completely new from the database:
// reattach an object from the database, only if it is not already part of it if (OBDal.getInstance().getSession().contains(myObject)) { OBDal.getInstance().getSession().lock(myObject.getEntity().getName(), myObject, LockMode.NONE); } // or reread it again, this example assumes that the object // is a Business Partner BusinessPartner businessPartner = OBDal.getInstance().get(BusinessPartner.class, myPreviouslyReadBP.getId());
When choosing the 'lock' strategy it is important to choose the correct lock order. First lock the objects which are used by the other objects.
ERROR org.hibernate.LazyInitializationException - could not initialize proxy - no Session
Hibernate will lazily load objects which are referenced from other objects. So when an object A references an object B, then when reading A from the database, hibernate will create a so-called proxy object B, which is used by A. The B proxy object is not initialized (its data is not set), this initialization is done when B is accessed directly (calling a method on it for example). To do this initialization hibernate needs the original hibernate session which was used to read A and create the B proxy object.
The LazyInitializationException occurs when this session is not valid anymore. This happens for example when OBDal.getInstance().commitAndClose() is called, this will close and remove the session. If an uninitialized proxy object is accessed after the commitAndClose call then this exception is thrown.
To prevent this make sure that the objects that you need after the commitAndClose are all initialized. This can be done using the Hibernate.initialize(object) call or by calling an arbitrary method on the object (for example getIdentifier()).
REST Web Services
I am getting: 'Caused by: Object Entity(null) is new but not writable'
The REST webservice uses the default role, client and organization of the user. The error Caused by: Object Order(null) is new but not writable is caused by the fact that either the role or the organization of the user do not allow the writing/creating the entity you are passing in.
See this topic in the Openbravo forum.
And see this issue for a related bug fix.
My REST webservice does not return anything
If your REST webservice only returns this:<ob:Openbravo/>
Then most of the time the user you used to log in is not authorised to see the requested information. Another reason for not seeing information is that the system will automatically add client and organization filtering to the REST query and only show objects which have active==true.
Using the REST XML Schema to generate (de-)serialization code
The REST XML Schema can be used to generate java or C# code which can be used to (de-)serialize the REST xml content in a type safe way. One thing to be aware of is that primitive typed elements (such as a name or description) have attributes defined in the XML Schema. This means that code generators will not create a String (or other primitive type) member/accessor but will instead use a separate generated type to encapsulate the primitive type value. This separately generated type also holds the other attributes.
See this forum post for a discussion around this.
Implementing your own webservice
There is a howto on this topic here.
Firefox Add-On:Poster, to test webservices
To support web service testing, Firefox has a nice add-on: Poster. This add-on allows you to POST/PUT XML to a URL and do GET and DELETE requests.
Application Dictionary Data Model
Using search references not to show a drop down in non-editable fields
For the complete reference about Search reference, read the Search section in Data Model document.
A Search reference does not show a drop down list but just text box with the selected value and a button to call the search next to it. When the field is non writable the button is not shown and only appears the textbox.
If a column is known to be used always in non-editable fields and it follows the naming rule defined for TableDir references. It is a good practice not to set it as TableDir reference but as Search one, in this case is not necessary to set the Reference Search Key field. Doing this the generated page will be lighter because instead of generating a drop down list with all the elements it will just generate a textbox.
Application Dictionary syntax
Application Dictionary contains many examples of default values, display logics, read-only logics and validations syntax. Just query your database to find out how they are defined (you can execute these queries inside Openbravo ERP, logged with an user with System Administrator role inside Application Dictionary || Maintenance || SQL Query). Find more info about dynamic expressions in the Dynamic Expressions chapter of the Developer's Guide.
Default values
SELECT t.name AS table_name, c.name AS column_name, c.defaultvalue AS default_value FROM ad_table t, ad_column c WHERE c.defaultvalue IS NOT NULL --AND c.defaultvalue like '%+%' --AND c.defaultvalue like '%SQL%' AND c.ad_table_id = t.ad_table_id ORDER BY t.name, c.name;
Display logics
SELECT w.name AS window, t.name AS tab, f.name AS FIELD, f.displaylogic AS display_logic FROM ad_window w, ad_tab t, ad_field f WHERE f.displaylogic IS NOT NULL AND w.ad_window_id = t.ad_window_id AND t.ad_tab_id = f.ad_tab_id ORDER BY w.name, t.name, f.name;
Example 1 If you want to restrict the access to a child tab based on the value of a field of its father tab you have to:
- Mark the column of the field inside the father's tab as Stored in Session.
- Add to all the fields of the child tab a display logic in the form @Father_Column_Name@ = 'Value'
See, for instance, how it is done in Master Data Management || Product Setup || Attribute || Attribute >> Attribute Value when un-checking List check-box in Attribute tab by querying:
SELECT w.name AS window, t.name AS tab, f.name AS FIELD, f.displaylogic AS display_logic FROM ad_window w, ad_tab t, ad_field f WHERE f.displaylogic IS NOT NULL AND w.ad_window_id = t.ad_window_id AND t.ad_tab_id = f.ad_tab_id AND UPPER(t.name) LIKE '%ATTRIBUTE VALUE%' ORDER BY w.name, t.name, f.name;
Read-only logics
SELECT t.name AS table_name, c.name AS column_name, c.readonlylogic AS read_only_logic FROM ad_table t, ad_column c WHERE c.readonlylogic IS NOT NULL AND c.ad_table_id = t.ad_table_id ORDER BY t.name, c.name;
Validations
SELECT t.name AS table_name, c.name AS column_name, vr.code AS validation_rule FROM ad_val_rule vr, ad_table t, ad_column c WHERE c.ad_val_rule_id = vr.ad_val_rule_id AND c.ad_table_id = t.ad_table_id ORDER BY t.name, c.name;
Build tasks
After export.database all the xml headers use double quotes
After export.database you see this in the top of the header:
<?xml version="1.0" encoding="UTF-8"?>
when diffing this with the repository you see that the repository has this (uses single quotes):
<?xml version='1.0' encoding='UTF-8'?>
The standard in Openbravo is to use single quotes. To solve this issue download stax-api-1.0.1.jar (StAX) and wstx-asl-3.0.2.jar (Woodstox) XML processors. Then, copy these two files to lib folder inside your Ant folder.
In a development environment:
Copy stax-api-1.0.1.jar and wstx-asl-3.0.2.jar files from Ant lib folder to Eclipse IDE lib folder (usually eclipse/plugins/org.apache.ant/lib/). Then, go to Eclipse IDE Window > Preferences... menu element and select Ant > Runtime > Ant Home Entries (Default). Click on Add External JARs... button and add stax-api-1.0.1.jar and wstx-asl-3.0.2.jar.
| Languages: |
ERP 2.50:Developers Guide/Examples/Reference | ERP 2.50:Developers Guide/Best Practices