Devart Forums

HCRoman

Hi devart team,

Any new information available for this problem???

Roman

HCRoman

Hi devart team,

the 4.2.110 version of EntityDeveloper does not update added procedure parameters for an existing procedure if I use update model from database.

The added parameter is shown as difference and marked for added, but nothing happens after finish updating the model..

Roman

HCRoman

Hi devart team,

where can I change and preset the default settings for the integrated and standalone EntityDeveloper?

Every time I generate a new model from scratch (database first), I have to change a couple of setting switches to my preferrences,
again and again...

(After this, the settings are stored in the edmx, thats great...., but not for the next new model).
I like Capitalized, Unchanged, no schema name etc.

Is there a edmx template or an edmx.default or can I do this in the config or registry??

Thanks for any advice

Roman

HCRoman

Shalex wrote:
HCRoman wrote:if you do not use merge, couse of some missing features and create a new model from database twice, you will get two edml files with all ed:Guid are different from the previous generated file! Why?
This is a designed behaviour. ed:Guid is our internal attribute and should identify uniquely each element on the model. How are you using ed:Guid?

I do not "use" it, but if you use WinMerge for instance, to "compare" old and new edml, you will have tons of "differences" hiding the real changes.
I use WinMerge to control for unwanted changes in my large model...

Shalex wrote:
HCRoman wrote:If you want to preserve really all user changes on names and column data types and table splitting, you have to mark and copy each user changed attribute in your edps file as modified, save the original generated in edps and during "merge" you can compare original versions and can mark it only as a change if the original has changed compared to the new generated one.
As a good designer help, you could color all "user changes", so you can find them easier.
Sorry, but it is not clear enough. Could you please give us an example?

Ok, you have an table TEST under oracle database with column VALID as NUMBER(1) and column FLAG as NUMBER(1), both are imported as BOOL by default. You change the second column to INT (int32) because it is not a bool!

Save and merge the model again and now you will see the second colum as "changed", but it is not the case. If you would have saved the "original" imported BOOL data type in the edps than you could see "nothing has changed" on that table column since last import.

Same could be done for any user "renames" of associations, entity sets or column names..

Is this explained enough?

Roman

HCRoman

Hi devart team,

if you do not use merge, couse of some missing features and create a new model from database twice, you will get two edml files with all ed:Guid are different from the previous generated file! Why?

try it out.

If you want to preserve really all user changes on names and column data types and table splitting, you have to mark and copy each user changed attribute in your edps file as modified, save the original generated in edps and during "merge" you can compare original versions and can mark it only as a change if the original has changed compared to the new generated one.
As a good designer help, you could color all "user changes", so you can find them easier.

For example in oracle we have a lot of NUMBER(1) columns and you can not distinguish if it is a bool or really a number(1), so you have to change this in the model...

Roman

HCRoman

Hallo devart team,

it seems to be very hard to fix this error

Now every time you change the file attribute outside or during save with overwrite you get the message that the file should be reloded. Before and after that, you can edit and save, but no changes are written!
As before, you have first to change the edml-file from readonly to readwrite and than you should open it in VS2010 designer...

Hope for next version!

Roman

HCRoman

Hi devart,

the new version of Entity-Developer
build 6.50.244 / 4.1.71
hase some new pluralization "feature"??

The SSDL names imported from Oracle are every time uppercase and the pluralization extensions generated by devart are every time lower case.

In the new version it comes mixed, some times the old behav and other names with all upper case.

It seems to depend on the last character from the original name,
if it is a digit or an vowel, all is pluralized now to upper case, else the pluralization is as old behav lower case????

---------
(This is as long as the Guid exists in the edml)

Second thing ist the ugly thing, that all internal ed:Guid changes every time you change/import a model. This does not help to find out what has really changed.....
Why this internal number is not deterministic calculated and will only change if the property has changed???

--------------

Why you change the naming for the *Designer.cs to Designer 1.cs. No please not!!!

Why the Model.info file is now included to the project??? No please not!

Please do not change such basic things with every version!!!

Let the user decide how to name the files, diagrams, view-files etc. and do not change this, please.

Any help or idea for all topics

Tnx

Roman

HCRoman

Hi devart team,

the test project was sent to your email address...

Roman

HCRoman

Hi devart team,
in version 6.30 you anounced for LinqConnect
...
Performance of ordered queries with the Skip() and Take() methods is improved: the 'rownum' pseudo-column is now used instead of the 'row_number' function

is this also possible for dotConnect for Oracle

The performance of the generated SQL for Entity-Framework is verry poor under some circumstances (JOINS in the underlying query)

SELECT
...
row_number() OVER (ORDER BY "Extent1".BETRIEBSSTAETTEID ASC) AS "row_number" FROM BETRIEBSSTAETTE_V02 "Extent1"
WHERE "Extent1".QUARTAL = 20114
ORDER BY "Extent1".BETRIEBSSTAETTEID ASC
) "Filter1"
WHERE "Filter1"."row_number" > :skip AND ROWNUM :skip AND ROWNUM <= :take
ORDER BY "Filter1".BETRIEBSSTAETTEID ASC

The performance is acceptable < 1 second

Any Help?

Roman

HCRoman

Hi Shalex,

this is not really the reason...

I every time set the correct max length to varchar parameters.
If the parameters length is less than 4000, the error occurres only later...

there is definitly a memory leak in the code, may be it is not in the devart dll. The memory is freed after closing and disposing the connection!
See my previous post.

Devart-Batch Updates are not a good solution, up to this release.
It does not support returning clause, so if you have generated keys or computed values, you can not work without hurts.

My ObjectContext Extension for the entity framework can insert, update and delete using ExecuteArray. Insert and Update works with correct returning clauses for all identity and computed columns.
So all references updated correctly after AcceptChanges for the EntityState of the Entity.

You have only to ensure the correct order of the operation.

Update and delete can also optionally use different logical keys for the operation...

Include the calls in the SavingChanges Event it ist mostly transparent for the user but hundred times faster... than SaveChanges

Roman

HCRoman

I have found a usable workaround for this....

(But it would be better, if the problem could be fixed)

Code: Select all

using (var tran = new TransactionScope(TransactionScopeOption.Required, TimeSpan.FromMinutes(25)))
{
  var nextJunk = true;
  while(nextJunk)
  {
    create object context.

    create a junk of data that does not throw an OutOfMemoryException.

    do ArrayInsert

    commit context, if needed

    dispose ObjectContext and ObjectContext.[b]Connection[/b]
    (the connection holds that data cousing the OutOfMemoryException)

    nextJunk = more data to insert...
  }
  tran.Complete();
}

(
So I can insert millions of records via entity framework in a couple of hours

)
Roman

HCRoman

Hi devart team,

I have the same problem, in OCI mode the Devart api throws an OutOfMemory-Exception after about 64000 records.

This seems to be an memory leak problem!!
(or fragmentation of large object heap???)

If I have two varchar 4000 columns, the exception occures after 64000 rows

If I shrink this two columns to 400, the exception occures only later....

Version:
5.0.284.0 / 6.30.185.0

In Direct mode all works fine.

Why this error is not fixed????

(Direct mode has also some limitations...)

We have to use windows authentification!
for our database connections

this is one limitation of devart direct mode

second limitation:

RETURNING clause does also not work!!!

so direct mode is not a good workaround for me.

(TOAD does this direct mode connection as external user with windows authentification!)

Any help available

Thanks

Roman

HCRoman

Hi devart team,

we are successful using devart EF4 with WPF clients.

Now i want to write a silverlight web app, that uses the existing EF4 entity model. I started with VS2010 BusinessApplication-Template and changed the web.config, so it should use devart oracle entity connection, but without success.

I moved the connectionString from my wpf app.config to the web.config and get the error message that metadata is an invalid connection string parameter...

Can you provide a simple example based on that BusinessApplication-Template but with an EntityModel residing in its own dll but embedded metadata in the Web server application.

(The Sample folder does not give me the clue, I am starting the usage of silverlight...)

Have I to include all classes from InstallWebTables.sql into my entity model? ore is it better to have two connections, one for handling user logon and one entity context connection for application stuff?

Show, how to work without connection string in the web.config and EXTERNAL user logon for oracle.

Spent some wise words about licensing, library referencing in the web.config and embedding metadata in the application.

(Shipping this as an VS2010 template in future releases would be great...)

Thank you for any help

Roman

HCRoman

Better is to track the StateChange event
This works for each connection open!

(
The code inside of the event handler can be one of the examples from Shalex too...
)

Code: Select all

partial void OnContextCreated() 
{ 
    Connection.StateChange += this.EntityConnection_StateChange;
} 

private void EntityConnection_StateChange(object sender, StateChangeEventArgs e)
{
   if (e.CurrentState == ConnectionState.Open)
   {
      DbConnection connection = ((EntityConnection)sender).StoreConnection;
      using (DbCommand command = connection.CreateCommand())
      {
         this.SetRoles(command);
         this.SetCurrentSchema(command);
      }
   }
}

Roman

HCRoman

Hi devart team,
after checking the new version, the edml is overwritten

but

with View generation enabled and read only PregeneratedViews.cs it still failes

// Custom tool 'DevartEfGenerator' failed.
// - Der Zugriff auf den Pfad "RlvModel.PregeneratedViews.cs" wurde verweigert.

Hope for next version...

Roman

Devart Forums

Search found 55 matches

Any new information available for this problem???

Update from Database does not add new procedure parameters

Default model settings possible?

Re: GUIIDs recreated with every new creation from database

GUIIDs recreated with every new creation from database

Opening .edml file causes it to be 'edited' with no changes'

Entity-Developer generates mixed lower uppercase SSDL names

Test-Project was sent by email....

Performance of ordered queries with the Skip() and Take()

This is not the reason...

workaround found..

Same problem, why is this not fixed in the last version???

Silverlight sample for VS2010

better use StateChange to set schema and roles...??