[转]Porting to Oracle with Entity Framework NLog

本文转自：http://izzydev.net/.net/oracle/entityframework/2017/02/01/Porting-to-Oracle-with-Entity-Framework.html

We had to port our giant-_ish_, Entity Framework based application to work on Oracle RDBMS (11g, later 12c).

We did it.

And we learned a lot.

Boss comes around and…

One day boss comes around and says:

Boss: “We need our application to work on Oracle”

Dev: “Oracle? But we use MS SQL…”

Boss: Can you do it?

Dev: [thinking at loud] We use Entity Framework, which is somewhat abstraction on top of our database. I’ve heared that there are some providers enabling work with Oracle, but never used them. I’ve seen this Microsoft conference where leading programmer was swapping SQL Server provider with Oracle one by a single line and it worked like a charm, but it was template-basic application and…

Boss: Great! We need this in 6 weeks.

Dev: But…

Boss: Inform me constantly. It’s very important project and fail is not an option. Don’t forget we need some time for tests. 2 weeks should be enough I guess. I gotta go now, I have another meeting. Thanks !

* Sound of the closing door. *

Dev: …what just happened?

Accept and advance

This is fromatted versions of all information gathered by 3 folks working on Oracle port for couple weeks.

I believe we wouldn’t do it if would not cooperate and solve each problem together.

All problems we’ve encountered were put on our slack channel with short solution and decryption.

It was the best decision we made in whole project.

Most common problems

Note: If you have troubles that are not covered in this article, here’s some notes from Oracle. Some of the issues are described there as well as are here. THOSE ARE THE MOST VALUABLE LINKS IN WHOLE ARTICLE.

If you google more you will find other resources on Oracle servers. Unfortunately, I discovered them when I was near finishing writing whole document.

Those files can be found inside different ODP.NET nuget packages (those readme files you never read).

This is what happen when you don’t read readme files.

Which provider should I use?

We recommend using Official Oracle ODP.NET. All cases presented here use this and only this provider. Version is 12.1.2400.

Dependencies
EntityFramework (>= 6.0.0 && < 7.0.0)
Oracle.ManagedDataAccess (>= 12.1.2400 && < 12.2.0)

Schema == User

This is the first thing you should learn about Oracle. In Oracle, there are no databases. There are schemas and schemas are users.

Do you want to create new database on your localhost like you did on MS SQL? Create new user. Its name is the name of your datab… I mean schema.

You can connect to your new schema with new connection in Oracle SQL Developer. Remember to assign to the created user Connect role (and maybe some others, I used to grant it with all roles, since I didn’t have to deal with security issues).

Note: User system has Other users. Those are all users on your Oracle server. You can access their database objects (dbo) from system user level, without connecting to the user itself. It might come handy in scenarios when you often repeat drop/create users.

DBO names.Length <= 30

All table names in your schemas must contain less or equal than 30 characters.

To configure your entity to map to table with specified name use ToTable(:string) in your EntityTypeConfiguration<> derived class.

public class FoobarEntityConfiguration : EntityTypeConfiguration<Foobar>
{
  public FoobarEntityConfiguration()
  {
    ToTable("Foobar");

    // More configuration here ...
  }
}

http://stackoverflow.com/a/756569/864968

I’d suggest creating convention test that checks every oracle table configuration defines table name no greater then 30 characters.

EF logger

This is extremely helpful when a) you don’t have Oracle license (which contains profiler and developer edition not) and b) you want to peek what’s going on under the hood after your C# is magically transformed to some kind of SQL.

EF allows to log executed queries.

private void EnableDebugLogs(DbContext context)
{
  context.Database.Log = s => Debug.WriteLine(s); // SQL writer
}

Logged queries can be found in Output window in Visual Studio. You can provide any action particularly.

ORA-01918: user ‘dbo’ does not exist

You’ve created EF migration, run Update-Database and you get this error.

Remember what I was talking about schemas and users? They’re the same thing!

In DbContext derived class specify schema name. It should be the same as the user you’re connecting with database.

public class FooDbContext : DbMigrationsConfiguration<FooDbContext>
{
  protected override void OnModelCreating(DbModelBuilder modelBuilder)
  {
    modelBuilder.HasDefaultSchema(" < Your connection string user here > ".ToUpper()); // Make sure it‘s upper case !
  }
}

ORA-01005: null password given; logon denied

This might happen while you’re running Update-Database and your ConnectionString does not contain Persist Security Info.

http://stackoverflow.com/questions/14810868/ora-01005-null-password-given-logon-denied

Migration directory

If you, just like us, have to support both MS SQL and Oracle, you should consider seperate migrations set for each RDBMS. To do so, configure your migration directory in DbMigrationConfiguration derived class.

public class FooMigrationConfiguration : DbMigrationsConfiguration<FooContext>
{
  public FooMigrationConfiguration()
  {
	DatabaseHelper.SetMigrationDirectory(this, " i.e. Contexts\Foo\Migrations  ");
  }
}

public static void SetMigrationDirectory<TContext>(DbMigrationsConfiguration<TContext> migration, string migrationsPath) where TContext : DbContext
{
  if (string.IsNullOrEmpty(migrationsPath))
  {
    throw new ArgumentException($"{nameof(migrationsPath)} cannot be null or empty.");
  }

  if (migrationsPath.Contains("/"))
    throw new ArgumentException($"Invalid {nameof(migrationsPath)}. Path should be valid Windows path. Use backslashes instead of slashes.");

  migration.MigrationsDirectory = migrationsPath;
}

Note: Speaking aside, support for both RDBMS is like maintaining two similar but different applications. They just differ. It is very unlikely that you will end up with application working on both MS SQL and Oracle without any changes.

EF is great tool, but don’t expect it to do miracles.

NLog + Oracle

If you used to log exception errors directly to Database with help of NLog then you should also modify your database target in your nlog.config.

<!--NLog configuration-->
<!--For more info visit: https://github.com/NLog/NLog website -->
<nlog xmlns="http://www.nlog-project.org/schemas/NLog.xsd"
      xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
      xsi:schemaLocation="http://www.nlog-project.org/schemas/NLog.xsd NLog.xsd"
      autoReload="true"
      throwExceptions="true"
      internalLogLevel="Off"
      internalLogFile="App_Data\nlog-internal.log" >

  <targets>

    <!--Puts logs into database-->
    <target name="db" xsi:type="Database" commandType="Text" connectionStringName="FoobarDbContext">
      <commandText>
        insert into FOOBARSCHEMA."FOOBARTABLE"("Id", "Level", "Logger", "Message", "StackTrace", "Date", "TenantId", "UserId")
        values(sys_guid(), :LogLevel, :Logger, :Message, :StackTrace, systimestamp, :TenantId, :UserId)
      </commandText>

      <!--Reserved words by Oracle (don‘t use them as variable names: Level, Date-->
      <parameter name="LogLevel" layout="${level}"/>
      <parameter name="Logger" layout="${logger}"/>
      <parameter name="Message" layout="${message}"/>
      <parameter name="StackTrace" layout="${exception:format=Message,Type,Method,StackTrace,Data:separator=\r\n\r\n:maxInnerExceptionLevel=10:innerFormat=Message,Type,Method,StackTrace,Data:innerExceptionSeparator=\r\n\r\n}"/>
      <parameter name="TenantId" layout="${event-properties:item=TenantId}"/>
      <parameter name="UserId" layout="${event-properties:item=UserId}"/>

    </target>
  </targets>

  <rules>
    <logger name="*" minlevel="Warn" writeTo="trace,db" />
  </rules>

</nlog>

Web.config

It’s quite clear that we should maintain at least two versions of configuration file.

We decided to name them accordingly Web.Oracle.config and Web.MSSQL.config. It’s clear and simple. During deploy the correct file is being selected and included into artifact.

You can also write watcher that detects Web.config change and accordingly replaces rest of web.configs in whole solution. You can do that like we did, with help of Cake Watch.

External app.settings section

It is a good practice to have separate config file with common settings for both MS SQL and Oracle version. We can exclude them by adding file attribute.

<appSettings file="AppSettings.config" />

Connection strings

Connections string differ so much that should be kept separately.

<connectionStrings>   

  <add name="ApplicationDbContext"
       providerName="Oracle.ManagedDataAccess.Client"
       connectionString="User Id=ApplicationDbContext;Password=P4S5W0RD;Data Source=OracleDataSource;Persist Security Info=true" />

</connectionStrings>

<entityFramework>
  <defaultConnectionFactory type="Oracle.ManagedDataAccess.EntityFramework.OracleConnectionFactory, Oracle.ManagedDataAccess.EntityFramework" />

  <providers>
    <provider invariantName="Oracle.ManagedDataAccess.Client"
              type="Oracle.ManagedDataAccess.EntityFramework.EFOracleProviderServices, Oracle.ManagedDataAccess.EntityFramework, Version=6.121.2.0, Culture=neutral, PublicKeyToken=89b483f429c47342" />
  </providers>
</entityFramework>

<system.data>
  <DbProviderFactories>
    <remove invariant="Oracle.ManagedDataAccess.Client" />
    <add name="ODP.NET, Managed Driver"
         invariant="Oracle.ManagedDataAccess.Client"
         description="Oracle Data Provider for .NET, Managed Driver"
         type="Oracle.ManagedDataAccess.Client.OracleClientFactory, Oracle.ManagedDataAccess, Version=4.121.2.0, Culture=neutral, PublicKeyToken=89b483f429c47342" />
  </DbProviderFactories>
</system.data>

<oracle.manageddataaccess.client>
  <version number="*">
  <dataSources>
  <dataSource alias="OracleDataSource"
              descriptor="(DESCRIPTION=(ADDRESS=(PROTOCOL=tcp)(HOST=localhost)(PORT=1521))(CONNECT_DATA=(SERVICE_NAME=ORC)))" />
  </dataSources>
  </version>
</oracle.manageddataaccess.client>

ODP.NET GAC (Global Assembly Cache)

Uninstall

Sometime you might get following error

Could not load type ‘OracleInternal.Common.ConfigBaseClass‘ from assembly ‘Oracle.ManagedDataAccess, Version=4.121.2.0, Culture=neutral, PublicKeyToken=89b483f429c47342‘.

This is not clear to me. Sometime I have to uninstall it, sometime I have to install it (due the error suggests something opposite).

I presume you have gacutil in your PATH. To check out, simply type gacutil /? to see available parameters. If you don’t, google for gacutil location and add it to your system variable PATH, or use gacutil with absolute path. Remember to open command window in Administrator mode.

gacutil /u Oracle.ManagedDataAccess

C:\...\packages\Oracle.ManagedDataAccess.12.1.2400\bin\x64
λ gacutil /u Oracle.ManagedDataAccess
Microsoft (R) .NET Global Assembly Cache Utility.  Version 4.0.30319.33440
Copyright (c) Microsoft Corporation.  All rights reserved.

Assembly: Oracle.ManagedDataAccess, Version=4.121.2.0, Culture=neutral, PublicKeyToken=89b483f429c47342, processorArchitec
ture=MSIL
Uninstalled: Oracle.ManagedDataAccess, Version=4.121.2.0, Culture=neutral, PublicKeyToken=89b483f429c47342, processorArchi
tecture=MSIL
Number of assemblies uninstalled = 1
Number of failures = 0

After that, reset IIS with iisreset to reload loaded assemblies and reload the page.

ref: Oracle .Net ManagedDataAccess Error: Could not load type ‘OracleInternal.Common.ConfigBaseClass’ from assembly

Install

Type is not resolved for member ‘Oracle.ManagedDataAccess.Client.OracleException,Oracle.ManagedDataAccess, Version=4.121.2.0, Culture=neutral, PublicKeyToken=89b483f429c47342‘.

Generally speaking all you got to do is to install Oracle.ManagedDataAccess.dll to your GAC (Global Assembly Cache).

gacutil /i <dll_path>

Oracle.ManagedDataAccess.dll can be found in your nuget package lib directory (mine was .\packages\Oracle.ManagedDataAccess.12.1.2400\lib\net40).

C:\...\packages\Oracle.ManagedDataAccess.12.1.2400\lib\net40> gacutil /i .\Oracle.ManagedDataAccess.dll
Microsoft (R) .NET Global Assembly Cache Utility.  Version 4.0.30319.33440
Copyright (c) Microsoft Corporation.  All rights reserved.

Assembly successfully added to the cache

ref: Entity Framework Seed method exception

Bounty

Install and uninstall enable further work, but what really would do the trick is a proper configuration that resolves constant installing and uninstalling assemblies from cache. So if come up with any better solution, feel free to share.

Identity

In both 11g and 12c, GUID Identity columns in migration files must be replaced from

Id = c.Guid(nullable: false, identity: true),

to

Id = c.Guid(nullable: false, identity: false, defaultValueSql: "SYS_GUID()"),

This is for both 11g and 12c

Identity in 11g

11g does not offer autoincrement + uniquness feature (commonly known as Identity in MS SQL). EF handles generating next Identity values by incrementing sequence by a trigger. Sequences are good for numeric column types, but they don’t work well with GUIDs.

Change generated triggers to insert SYS_GUID() (which is NEWID() equivalent in MS SQL) or change C# migration.

Apply

"Oracle 11.2.0.2.0 does not support APPLY" exception

Error says everything. More robust EF queries are not supported on 11g. You can rewrite your query, but it’s just workaround, not a solution for this particular problem. Apply are supported from version 12c. This is the main reason we moved from 11g to 12c. Fortunately, client was moving to 12c as well.

ref: ODAC 11.2 Release 4 (11.2.0.3.0) throwing “Oracle 11.2.0.2.0 does not support APPLY” exception

Bulk inserts

Bulk inserts are possible with ODP.NET Unmanaged Driver.

Isolation levels

This one is quite important. In EF we can create transaction with specified Isolation level. Those differ in MS SQL and Oracle.

MS SQL 2014 isolation levels

  • READ UNCOMMITTED
  • READ COMMITTED
  • REPEATABLE READ
  • SNAPSHOT
  • SERIALIZABLE

Oracle 11g/12c isolation levels

  • READ UNCOMMITTED
  • READ COMMITTED
  • REPEATABLE READ
  • SERIALIZABLE

I’m no Oracle expert, so if you want gain more knowledge about those, check out this awesome blog post. It helped me a lot in understanding differences in isolation levels between Oracle and MS SQL.

High Oracle Memory usage

High RAM Memory Consumed by Oracle 11g for Windows Server 2008

PL/SQL tips

clear screen;

SET FEEDBACK OFF;

/ and ; When do I need to use a semicolon vs a slash in Oracle SQL?

commit; commits changes on database. You have to put this after every inserting section in your script.

SID vs SERVICE_NAME

How SID is different from Service name in Oracle tnsnames.ora

Quartz.NET + Oracle

Some configuration need to be done before your scheduler will run on Oracle.

jobStore.driverDelegateType

properties["quartz.jobStore.driverDelegateType"] = "Quartz.Impl.AdoJobStore.OracleDelegate, Quartz";

http://stackoverflow.com/a/20343752/864968

dataSource.default.provider

["quartz.dataSource.default.provider"] = "OracleODPManaged-1211-40",

http://www.quartz-scheduler.net/documentation/quartz-2.x/tutorial/job-stores.html

Migrations

It is highly possible that you’re gonna mess around with migrations, DbContexts and entities. You might end up with changes in your model that will need new migration.

We had one case when after adding a migration (for MSSQL) we got a lot of table creates like this migration was the very first, but it wasn’t.

To avoid such issues with your model changes (when you’re absolutly sure there were none, like any table definition changes), add ContextKey and MigrationsNamespace to your MigrationConfiguration.

ContextKey is stored in __MigrationHistory table, created by EF migrator during first successful migration.

MigrationsNamespace is a bit different. It will appear in C# migrations. During migration on database those migrations transformed to xml (edmx), gzipped, compressed with “base64” and finally stored in __MigrationHistory table. This will cause differences in model.

Also, correct namespace is crucial while looking for proper migrations during EF migrations scan.

Compressed model can be easly decompressed. Check this link for more.

Can I get decode an EntityFramework Model from a specified migration?,

Entity Framework Migrations Rebuild Target Hash

More here:

Changing the Namespace With Entity Framework 6.0 Code First Databases,

Namespace changes with Entity Framework 6 migrations

Create table if does not exist

The easiest way is to handle it with try/catch section.


begin
  execute immediate
  ‘
    create table "SCHEMA"."__MigrationHistory"
    (
      "MigrationId" nvarchar2(150) not null,
      "ContextKey" nvarchar2(300) not null,
      "Model" blob not null,
      "ProductVersion" nvarchar2(32) not null,
      constraint "PK___MigrationHistory" primary key("MigrationId", "ContextKey")
    )
  ‘;
exception
  when others then
    if sqlcode <> -955 then
      raise;
    end if;
end;
/

source: http://stackoverflow.com/questions/15630771/check-table-exist-or-not-before-create-it-in-oracle

Drop all tables in schema

BEGIN
   FOR cur_rec IN (SELECT object_name, object_type
                     FROM user_objects
                    WHERE object_type IN
                             (‘TABLE‘,
                              ‘VIEW‘,
                              ‘PACKAGE‘,
                              ‘PROCEDURE‘,
                              ‘FUNCTION‘,
                              ‘SEQUENCE‘
                             ))
   LOOP
      BEGIN
         IF cur_rec.object_type = ‘TABLE‘
         THEN
            EXECUTE IMMEDIATE    ‘DROP ‘
                              || cur_rec.object_type
                              || ‘ "‘
                              || cur_rec.object_name
                              || ‘" CASCADE CONSTRAINTS‘;
         ELSE
            EXECUTE IMMEDIATE    ‘DROP ‘
                              || cur_rec.object_type
                              || ‘ "‘
                              || cur_rec.object_name
                              || ‘"‘;
         END IF;
      EXCEPTION
         WHEN OTHERS
         THEN
            DBMS_OUTPUT.put_line (   ‘FAILED: DROP ‘
                                  || cur_rec.object_type
                                  || ‘ "‘
                                  || cur_rec.object_name
                                  || ‘"‘
                                 );
      END;
   END LOOP;
END;
/

source: http://stackoverflow.com/a/1690419/864968

Inserting IDs to Identity columns

Sometime you have to insert row with specific ID. If your ID column is was defined with generated always as identity not null you will get the following exception: ORA-32795: cannot insert into a generated always identity column. This is the default way of generting scripts by EF btw.

You can replace this definition with genereted by default as identity on null. After doing this, you will be able to insert values to ID columns and nulls will be replaced with generated values.

Types mapping

Sometime we need to configure column types explicitly.

Extensions

We can use extensions.

String properties


using System;
using System.Data.Entity.ModelConfiguration.Configuration;

/// <summary>
/// String property configuration helpers.
/// More info about Oracle Data Provider mapping can be found here https://docs.oracle.com/cd/E63277_01/win.121/e63268/entityCodeFirst.htm
/// </summary>
public static class OracleStringPropertyConfigurationExtensions
{
  /// <summary>
  /// Configures string property for valid NVarChar2 column.
  /// Sets max length and encoding.
  /// </summary>
  /// <param name="property">Property to be configured.</param>
  /// <param name="maxLength">Column value max size. Size cannot be greater than 2000 bytes.</param>
  /// <returns>Property configuration</returns>
  public static StringPropertyConfiguration IsNVarChar2(this StringPropertyConfiguration property, int maxLength = NVARCHAR2_MAX)
  {
    if (property == null) throw new ArgumentNullException(nameof(property));
    if (maxLength > NVARCHAR2_MAX) throw new ArgumentException($"Oracle nvarchar2 column accepts strings that are not greater than {NVARCHAR2_MAX}.");

    property
      .HasMaxLength(maxLength)
      .IsUnicode(true);

    return property;
  }

  /// <summary>
  /// Configures string property for valid NVarChar2 column.
  /// Sets max length and encoding.
  /// </summary>
  /// <param name="property">Property to be configured.</param>
  /// <param name="maxLength">Column value max size. Size cannot be greater than 4000 bytes.</param>
  /// <returns>Property configuration</returns>
  public static StringPropertyConfiguration IsVarChar2(this StringPropertyConfiguration property, int maxLength = VARCHAR2_MAX)
  {
    if (property == null) throw new ArgumentNullException(nameof(property));
    if (maxLength > VARCHAR2_MAX) throw new ArgumentException($"Oracle varchar2 column accepts strings that are not greater than {VARCHAR2_MAX}.");

    property
      .HasMaxLength(maxLength)
      .IsUnicode(false);

    return property;
  }

  /// <summary>
  /// Configures string property for valid NVarChar2 column.
  /// Sets max length and encoding.
  /// </summary>
  /// <param name="property">Property to be configured.</param>
  /// <param name="maxLength">Column value max size. Size must be greater than 4000 bytes.</param>
  /// <returns>Property configuration</returns>
  public static StringPropertyConfiguration IsClob(this StringPropertyConfiguration property, int? maxLength = null)
  {
    if (property == null) throw new ArgumentNullException(nameof(property));
    if (maxLength < CLOB_MIN) throw new ArgumentException($"To configure column as CLOB type set its max length to be greater or equal than {CLOB_MIN}.");

    property.IsUnicode(false);

    // String Length
    if (maxLength.HasValue) property.HasMaxLength(maxLength);
    else property.IsMaxLength();

    return property;
  }

  /// <summary>
  /// Configures string property for valid NVarChar2 column.
  /// Sets max length and encoding.
  /// </summary>
  /// <param name="property">Property to be configured.</param>
  /// <param name="maxLength">Column value max size. Size must be greater than 4000 bytes.</param>
  /// <returns>Property configuration</returns>
  public static StringPropertyConfiguration IsNClob(this StringPropertyConfiguration property, int? maxLength = null)
  {
    if (property == null) throw new ArgumentNullException(nameof(property));
    if (maxLength < NCLOB_MIN) throw new ArgumentException($"To configure column as NCLOB type set its max length to be greater or equal than {NCLOB_MIN}.");

    property.IsUnicode(true);

    // String Length
    if (maxLength.HasValue) property.HasMaxLength(maxLength);
    else property.IsMaxLength();

    return property;
  }

  private const short NVARCHAR2_MAX = 2000;
  private const short VARCHAR2_MAX = 4000;
  private const short CLOB_MIN = 4001;
  private const short NCLOB_MIN = CLOB_MIN;
}

Binary properties


using System.Data.Entity.ModelConfiguration.Configuration;

/// <summary>
/// Binary property configuration helpers.
/// More info about Oracle Data Provider mapping can be found here https://docs.oracle.com/cd/E10405_01/appdev.120/e10379/ss_oracle_compared.htm
/// </summary>
public static class OracleBinaryPropertyConfigurationExtensions
{
  /// <summary>
  /// Configures property to map the type BLOB.
  /// BLOB is equivalent of MSSQL Image type.
  /// </summary>
  /// <param name="property">Property to be configured.</param>
  /// <returns>Property configuration</returns>
  public static BinaryPropertyConfiguration IsOracleImage(this BinaryPropertyConfiguration property)
  {
    if (property == null) throw new ArgumentNullException(nameof(property));

    property.HasColumnType("BLOB");

    return property;
  }
}

Numeric properties

using System.Data.Entity.ModelConfiguration.Configuration;

/// <summary>
/// Decimal property configuration helpers.
/// More info about Oracle Data Provider mapping can be found here https://docs.oracle.com/cd/E63277_01/win.121/e63268/entityCodeFirst.htm
/// </summary>
public static class OracleDecimalPropertyConfigurationExtensions
{
  /// <summary>
  /// Configures property to map the type Number with given scale and precision.
  /// https://docs.oracle.com/cd/E10405_01/appdev.120/e10379/ss_oracle_compared.htm
  /// </summary>
  /// <param name="property">Property to be configured.</param>
  /// <param name="precision">Number of digits in a number.</param>
  /// <param name="scale">Scale is the number of digits to the right of the decimal point in a number.</param>
  /// <returns>Property configuration</returns>
  public static DecimalPropertyConfiguration IsNumber(this DecimalPropertyConfiguration property, byte precision, byte scale)
  {
    if (property == null) throw new ArgumentNullException(nameof(property));

    if (precision < 1 || precision > 38)
      throw new InvalidOperationException("Precision must be between 1 and 38.");

    if (scale > precision)
      throw new InvalidOperationException("Scale must be between 0 and the Precision value.");

    property
      .HasColumnType("NUMBER")
      .HasPrecision(precision, scale);

    return property;
  }

  /// <summary>
  /// Configures property to map the type Number(19,4).
  /// This is equivalent of Money type from MS SQL.
  /// https://docs.oracle.com/cd/E10405_01/appdev.120/e10379/ss_oracle_compared.htm
  /// </summary>
  /// <param name="property">Property to be configured.</param>
  /// <returns>Property configuration</returns>
  public static DecimalPropertyConfiguration IsOracleMoney(this DecimalPropertyConfiguration property)
  {
    if (property == null) throw new ArgumentNullException(nameof(property));

    property.IsNumber(19, 4);

    return property;
  }
}

Consolidate

Keep both Oracle.ManagedDataAccess and Oracle.ManagedDataAccess.EntityFramework in the same version. It’s good practice to have assemblies of same version across whole all projects in the solution.

Example EF DbContext configuration

public class FoobarContext : DbContext, IFoobarContext
{
     ////////// DbSets //////////

    protected override void OnModelCreating(DbModelBuilder modelBuilder)
    {
        foreach (var configuration in GetConfigurations())
        {
            base.OnModelCreating(configuration.ConfigureModel(modelBuilder));
        }
    }

    protected virtual IEnumerable<IModelCreateConfiguration> ConfigurationsForMssql
    {
        get
        {
            yield return new MssqlModelCreateConfiguration();
            // Add more ‘yield return‘ for more CreateConfigurations
        }
    }

    protected virtual IEnumerable<IModelCreateConfiguration> ConfigurationsForOracle
    {
        get
        {
            yield return new OracleModelCreateConfiguration();
            // Add more ‘yield return‘ for more CreateConfigurations
        }
    }

    private IEnumerable<IModelCreateConfiguration> GetConfigurations()
    {
        return this.GetDatabaseType()
            .ReturnWhen(
                mssql: () => ConfigurationsForMssql,
                oracle: () => ConfigurationsForOracle);
    }
}

public interface IModelCreateConfiguration
{
    DbModelBuilder ConfigureModel(DbModelBuilder modelBuilder);
}

public class OracleFoobarModelCreateConfiguration : FoobarCreateModelConfiguration
{
    public override DbModelBuilder ConfigureModel(DbModelBuilder modelBuilder)
    {
        base.ConfigureModel(modelBuilder);

        modelBuilder.HasDefaultSchemaForOracle("Foobar", true);

        modelBuilder.Configurations.Add(new FooyConfigurationForOracle());
        modelBuilder.Configurations.Add(new BarConfigurationForOracle());
        modelBuilder.Configurations.Add(new BuzzConfigurationForOracle());

      	// more configurations ...

        return modelBuilder;
    }
}

public class OracleMigrationConfiguration : FoobarDbMigrationsConfiguration
{
    public OracleMigrationConfiguration()
        : base(new Arguments
        {
            MigrationDirectoryPath = @"Contexts\FoobarContext\Oracle\Migrations",
            MigrationNamespace = @"Foobar.FoobarContext.Oracle.Migrations"
        })
    {
    }
}

public abstract class MigrationConfigurationBase : DbMigrationsConfigurationBase<FoobarContext>
{
    protected MigrationConfiguration(Arguments arguments)
        : base(arguments.Configure(args =>
                         {
                             args.ContextKey = CONTEXT_KEY;
                         }))
    {
    }

    public const string CONTEXT_KEY = "FoobarContext";
}

public abstract class DbMigrationsConfigurationBase<TContext> : DbMigrationsConfiguration<TContext> where TContext : DbContext, new()
{
    protected FoobarDbMigrationsConfiguration(Arguments arguments)
    {
        DatabaseHelper.ConfigureAutomaticDatabaseMigration(this); // true
        DatabaseHelper.SetMigrationDirectory(this, arguments.MigrationDirectoryPath);
        DatabaseHelper.SetContextKey(this, arguments.ContextKey);
        DatabaseHelper.SetMigrationNamespace(this, arguments.MigrationNamespace);
    }

    /// <summary>
    /// Arguments for DbMigrationsConfiguration
    /// </summary>
    public class Arguments
    {
        /// <summary>
        /// Migration directory is the place where all migration classes files will be stored.
        /// </summary>
        public string MigrationDirectoryPath { get; set; }

        /// <summary>
        /// Context Key must remain the same so migration remain continous.
        /// </summary>
        public string ContextKey { get; set; }

        /// <summary>
        /// Generated migration classes namespaces.
        /// </summary>
        public string MigrationNamespace { get; set; }

    }
}

public static class DatabaseHelper
{
    public static void ConfigureAutomaticDatabaseMigration<TContext>(DbMigrationsConfiguration<TContext> migration) where TContext : DbContext
    {
        migration.AutomaticMigrationsEnabled = Convert.ToBoolean(WebConfigurationManager.AppSettings[SettingConstants.AutomaticMigrationsEnabled]);
        migration.AutomaticMigrationDataLossAllowed = Convert.ToBoolean(WebConfigurationManager.AppSettings[SettingConstants.AutomaticMigrationDataLossAllowed]);
    }

    public static void SetMigrationDirectory<TContext>(DbMigrationsConfiguration<TContext> migration,
                                                       string migrationsPath)
      where TContext : DbContext
    {
        if (string.IsNullOrEmpty(migrationsPath))
        {
            throw new ArgumentException($"{nameof(migrationsPath)} cannot be null or empty.");
        }

        if (migrationsPath.Contains("/"))
            throw new ArgumentException($"Invalid {nameof(migrationsPath)}. Path should be valid Windows path. Use backslashes instead of slashes.");

        migration.MigrationsDirectory = migrationsPath;
    }

    public static void SetContextKey<TContext>(DbMigrationsConfiguration<TContext> migration,
                                               string contextKey)
      where TContext : DbContext
    {
        if (contextKey != null)
        {
            migration.ContextKey = contextKey;
        }
    }

    public static void SetMigrationNamespace<TContext>(DbMigrationsConfiguration<TContext> migration,
                                                       string migrationNamespace)
      where TContext : DbContext
    {
        if (string.IsNullOrEmpty(migrationNamespace) == false)
        {
            migration.MigrationsNamespace = migrationNamespace;
        }
    }
}

public enum DatabaseType
{
    Mssql, Oracle
}

public static class DatabaseTypeProviderExtensions
{
    public static void PerformActionOn(this DatabaseType type,
                                       Action mssql,
                                       Action oracle)
    {
        switch (type)
        {
            case DatabaseType.Mssql:
                mssql();
                break;

            case DatabaseType.Oracle:
                oracle();
                break;

            default:
                throw new ArgumentOutOfRangeException();
        }
    }

    public static TResult ReturnOn<TResult>(this DatabaseType type,
                                            Func<TResult> mssql,
                                            Func<TResult> oracle)
    {
        switch (type)
        {
            case DatabaseType.Mssql:
                return mssql();

            case DatabaseType.Oracle:
                return oracle();

            default:
                throw new ArgumentOutOfRangeException();
        }
    }
}

public static class DbModelBuilderExtensions
{
    /// <summary>
    /// Set default schema name for context. Schema name must be valid with Oracle schemas standard.
    /// </summary>
    /// <param name="builder">Context model builder.</param>
    /// <param name="schemaName">Schema name to set.</param>
    /// <param name="capitalize">Capilize provided schema name.</param>
    /// <returns>Context model builder.</returns>
    public static DbModelBuilder HasDefaultSchemaForOracle(this DbModelBuilder builder,
                                                           string schemaName,
                                                           bool capitalize = false)
    {
        if (builder == null) throw new ArgumentNullException(nameof(builder));

        // Schema name validation

        if (string.IsNullOrWhiteSpace(schemaName))
            throw new ArgumentNullException(nameof(schemaName));

        if (schemaName.Length > 30)
            throw new ArgumentException("Schema name max length is 30.");

        if (capitalize)
        {
            schemaName = schemaName.ToUpperInvariant();
        }

        if (schemaName.All(char.IsUpper) == false) // This will also check for whitespaces ;)
            throw new ArgumentException("Schema name should contain no whitespaces and be uppercase.");

        builder.HasDefaultSchema(schemaName);

        return builder;
    }
}

/// <summary>
/// Enables fluent configuration.
/// </summary>
public static class ConfigureExtensions
{
    /// <summary>
    ///
    /// </summary>
    /// <typeparam name="T"></typeparam>
    /// <param name="system"></param>
    /// <param name="configure"></param>
    /// <returns></returns>
    public static T Configure<T>(this T system, Action<T> configure)
    {
        if (system == null)
            throw new ArgumentNullException(nameof(system));

        configure(system);
        return system;
    }
}

Disclaimer #1

I’m Oracle newbie, a .NET developer, challenged to achieve impossible with technology I used to have very little to do in the past. If you know any better solution, feel free to suggest or create Pull Request to this article. Thank you in advance!

Disclaimer #2

We’ve done it. We’ve delivered the product to client in the last day without any bigger bugs.

We were heroes.

Two weeks laters, product owners decided that we’re not merging oracle to master branch.

“We won’t have any clients with Oracle ever.”

时间： 2024-10-08 18:26:39

[转]Porting to Oracle with Entity Framework NLog的相关文章

Web Api 2, Oracle and Entity Framework

Web Api 2, Oracle and Entity Framework I spent about two days trying to figure out how to expose the Oracle.ManagedDataAccess 4.121.1.0 library over a Web Api 2.2 OData v4 Endpoint. Looking through the Oracle documentation, Oracle Managed Data Access

Oracle 与 entity framework 6 的配置，文档

官方文档: http://docs.oracle.com/cd/E56485_01/win.121/e55744/intro001.htm#ODPNT123 Oracle 对 微软 实体框架 EF6 的支持,在 ODP.NET 的新版本中才有实现. Oracle Data Access Components (ODAC)  Windows 下载:  ODAC 12c Release 3 包括支持  Entity Framework 6 Code First and Code First Migr

在Oracle中使用Entity Framework 6 CodeFirst

项目中需要将系统从SQLServer数据库迁移到Oracle上.由于原大部分数据访问操作都是通过包装了Entity Framework的统一访问入口实现的,所以需要研究Entity Framework从SQLServer转移到Oracle的实现方式. 自从EF4.X起,Oracle就为EF提供了驱动支持,但是很可惜的是不支持CodeFirst模式.庆幸的是从ODP.NET 11.2.0.3.0开始,Oracle官方提供了支持CodeFirst的纯托管代码的EF驱动.但是有以下几点是需要知道的;

Oracle中使用Entity Framework 6.x Code-First

Oracle中使用Entity Framework 6.x Code-First方式开发 去年写过一篇EF的简单学习笔记,当时EF还不支持Oracle的Code-First开发模式,今天无意又看了下Oracle官网,发现EF6.X已经支持了,并且给出了二篇教程(英文版): 1.Using NuGet to Install and Configure Oracle Data Provider for .NET 2.Entity Framework Code First and Code First

ODAC的安装以及Entity Framework for Oracle 基本配置

1.安装ODAC 根据自己操作系统x86,x64来判断下载的ODAC版本 http://www.oracle.com/technetwork/database/windows/downloads/utilsoft-087491.html 安装过程中,注意配置你所选择数据库服务器的位置和服务名 2.项目添加Entity Framework 此处连接地址注意修改,因为我自己是本地连接的,所以我当时服务器配置的是127.0.0.1,如果是远程连接,请在ODAC的安装目录下修改连接地址和相关信息(如下图

Entity Framework 处理MySQL,Oracle 并发问题

前提, 使用的 Entity Framework Power Tools 来生成Code First 模式 环境: EF6.0 MySQL 6.9 EF Power tools 4 VS2012 问题, SQL Server 使用 TS 字段来处理并发控制,  MySQL 如何处理 ? Oracle 如何处理? 思路:使用EF 的并发控制标识,加上数据库自动更新TS字段 1. 数据库自动更新TS, MySQL 可以设置TS列为 Timestamp 类型,加上默认值 NOT NULL DEFAUL

Entity Framework 学习

Entity Framework 学习初级篇1--EF基本概况... 2 Entity Framework 学习初级篇2--ObjectContext.ObjectQuery.ObjectStateEntry.ObjectStateManager类的介绍... 7 Entity Framework 学习初级篇3-- LINQ TOEntities. 10 Entity Framework 学习初级篇4--EntitySQL. 17 Entity Framework 学习初级篇5--ObjectQ

ASP.NET MVC with Entity Framework and CSS一书翻译系列文章之第一章：创建基本的MVC Web站点

在这一章中,我们将学习如何使用基架快速搭建和运行一个简单的Microsoft ASP.NET MVC Web站点.在我们马上投入学习和编码之前,我们首先了解一些有关ASP.NET MVC和Entity Framework的背景信息,然后再详细学习如何搭建开发环境. 1.1 MVC和ASP.NET MVC 这本书涵盖Microsoft版本的MVC,即ASP.NET MVC.在写本书的时候,ASP.NET MVC的生产版本为MVC5,因此,本书的示例代码都使用ASP.NET MVC 5编写.在本书中

[原创]Entity Framework查询原理

前言 Entity Framework的全称是ADO.NET Entity Framework,是微软开发的基于ADO.NET的ORM(Object/Relational Mapping)框架.Entity Framework的主要特点:1. 支持多种数据库(Microsoft SQL Server, Oracle, and DB2):2. 强劲的映射引擎,能很好地支持存储过程:3. 提供Visual Studio集成工具,进行可视化操作:4. 能够与ASP.NET, WPF, WCF, WCF