Devart Forums

Describing procedures on the context initialization was disabled in the release of dotConnect for Oracle 7. We recommend to download the latest 7.1.58 version. If you encounter any problems with it, please contact our Sales department (please specify the license number in the letter).

As for saving the schema name for package procedures, Entity Developer does so, e.g., for the 'Update from Database' wizard. To find and describe the procedure, the wizard has to properly parse its full name (i.e., determine whether the prefix is the schema or package name).

However, we will consider changing this behaviour if saving the name of the default schema is disabled.

Anyway, the "object does not exist" error should not occur in the latest version: no procedure describing is performed at the runtime, and when executing the procedure, its source is simply included into the command text (so that the Oracle server determines whether 'com_util' is a schema or a package on its own).

Please tell us if you can reproduce the issue when using the new version of dotConnect for Oracle.

We can only recommend you to use one of the following alternatives:

• negotiate with the hoster so that they either enable port forwarding or install your database on a trusted port;
• configure your application to be trusted (this needs the certain actions to be done by application users, so this solution will probably suit only if the application is supposed to work in intranet):
  http://msdn.microsoft.com/en-us/library/gg192793.aspx
• access the database from, e.g., a web service instead of working with it directly from the Silverlight application; in this case, you will have to switch from LinqConnect for Silverlight to, e.g., LinqConnect Professional.

We have fixed the problem with getting entities directly from the cache for commands like
'context.Depts.First(d => d.Deptno == literal)'.

Also, we have implemented getting entities by keys of type Int16.

These changes will be available in the nearest build. We will post here when this build is released.

According to the stack trace, the issue occurs when trying to describe a stored procedure. In previous versions of LinqConnect, Oracle procedures that return result sets were described when preparing metadata for the context. In the latest versions, the complete procedure metadata is stored in the mapping source (e.g., in the mapping attributes), thus instantiating the context should not result in the specified error. In particular, this means that with the newer versions you can change the schema before actually executing stored procedures.

JIC: besides that, the 'Initialization Command' connection string parameter is available since the 7.0 version of dotConnect for Oracle. This parameter allows specifying the command that will be executed just after opening the connection.

As for the error itself, please check the following:

• whether the stored procedures belong to the same user as the tables;
• whether the procedures are declared in a package;
• whether the user specified in the connection string is the same user to whom the procedures belong.

If possible, please send us the declarations of these procedures and/or the model you are working with, so that we are able to analyze the issue in more details.

The problem with Silverlight port limitations is a configuration one and is not actually related to the LinqConnect functionality.

As an alternative for changing the port your server uses, you can, e.g., configure forwarding from one of 'allowed' ports to 3306 on the machine where the MySQL server is installed. Port forwarding is discussed in details, e.g., here:
http://stackoverflow.com/questions/1476 ... g-bridging

Thank you for clarification, we have reproduced the issue. We will inform you when it is fixed.

Could you please specify the following:

• the type of the project with which the issue occurs (a console library, a WinForms/WPF application, an ASP.NET web site etc.);
• the language used in this application (VB or C#);
• whether any other documents are opened in Visual Studio when the problem occurs.

Thank you for the screenshot. We will analyze the situation and inform you about the results as soon as possible.

LinqConnect should work in the very same way as it is described in the provided article: if you are querying for a single entity by its primary key, LinqConnect searches for it in the entity cache at first. However, if there is no such entity in the cache, it has to query the database.

For example, consider the following code:
If you are observing some other behaviour, please describe the situation in more details.

Sorry for the delay.

It isn't possible to override the functionality of LinqConnect DataProviders. More importantly, querying disconnected DataSets is not the proper way to test the code that uses LinqConnect, at least because in this case the larger part of filtering/projecting logic is moved from the server to the client. (Since LinqConnect queries are translated into SQL, which is then executed in the database, they may give different results than 'the same' LINQ to Objects queries.)

Generally, we recommend to create a separated database specifically for testing purposes.

Please specify the following:

• whether you are using a stand-alone or Visual-Studio-integrated version of Entity Developer;
• in the latter case, please specify the edition and exact version of Visual Studio you are using;
• whether you installed any third-party products prior to the problem (if yes, please describe them).

If possible, please also try catching this exception and send us its details, including the stack trace. To do so, please perform the following:

• open your model;
• open a new Visual Studio instance;
• attach it to your process (the existing Visual Studio instance or stand-alone Entity Developer; the Tools -> 'Attach to Process' main menu item);
• ensure that you are catching all CLR exceptions (the Debug -> Exceptions main menu item, the 'Thrown' check box in the 'Common Language Runtime Exceptions' line);
• modify and save the model.

Could you please describe the scenario you are trying to implement in more details? For example, please specify the definitions of the database objects used and the differences between the initial 'Variety' set and the one you are trying to create (e.g., whether the entities in these sets have the same or different primary keys).

Also, I will send you a small test project (please check that it is not blocked by your mail filter). Please specify what should be changed in this sample or, if possible, send us your test project, so that we are able to reproduce and analyze the issue.

When the InsertOnSubmit method is invoked, no actual interoperations with the database are performed. The new entity is saved to the database only when SubmitChanges is called. Thus, the new row is not available in the database yet when you execute the query.

On the other hand, the query is translated into SQL and executed at the server. As the server returns an empty result set, the query result is null.

Thus, the 'target' object should not be null only if you invoke SubmitChanges before querying.

Please tell us if anything is unclear.

Thank you for the report. There is an issue with handling cursor parameters of stored procedures, and we are currently working on it. We will post here as soon as this problem is fixed.

The ObjectTrackingEnabled property specifies whether the context tracks the changes in the entity objects; i.e., whether it should subsctibe for the PropertyChanging event, create the entity copy storing the initial values etc. Regardless of the ObjectTrackingEnabled value, the context caches loaded entities.

The EntityCacheMode property affects the way the entities are stored in the cache: the context keeps either strong or weak references to them. By default, the WeakReference mode is used, allowing you to dispose your entity objects while the context is alive.

Please specify if anything is unclear.

Also, could you please describe the purpose of disabling entity caching in more details?