先说说 insert 吧.
当EF执行insert时，如果我们传入的对象是有关联(1对多等)的话，它会执行多个语句 insert到多个表，
并且再select出来填充我们的属性(因为有些column默认值是sql设定的，比如id等，我们insert后要有最新的数据丫).
using (EFDB db = new EFDB())
{
db.prods.Add(new Prod
{
code = "mk100",
name = "name",
detail = new ProdDetail //1对1关系当 prod_detail table
{
fullDescription = "des"
}
});
db.SaveChanges();
}
Opened connection at // :: +:
Started transaction at // :: +:
SET SESSION sql_mode='ANSI';INSERT INTO `prod`(
`code`,
`name`) VALUES (
@gp1,
@gp2);
SELECT
`id`
FROM `prod`
WHERE row_count() > AND `id`=last_insert_id()
-- @gp1: 'mk100' (Type = String, IsNullable = false, Size = )
-- @gp2: 'name' (Type = String, IsNullable = false, Size = )
-- Executing at // :: +:
-- Completed in ms with result: EFMySqlDataReader INSERT INTO `prod_detail`(
`prod_id`,
`id`,
`fullDescription`) VALUES (
,
,
@gp1)
-- @gp1: 'des' (Type = String, IsNullable = false, Size = )
-- Executing at // :: +:
-- Completed in ms with result: Committed transaction at // :: +:
Closed connection at // :: +:
Disposed transaction at // :: +:
delete 的话也有类似的情况，如果是关联的话，它也会执行多个语句去删除多个表，delete必须先把要洗掉的行select出来(这个似乎不太好...)
如果关联的数据是null,就不会去delete了，比如下面没有使用Include的话(虽然pk是有的，但是detail对象是Null)
WillCascadeOnDelete - Enabling Cascade Delete (级联删除) <-- 目前不清楚这样设置
using (EFDB db = new EFDB())
{
//db.Configuration.LazyLoadingEnabled = false;
db.prods.RemoveRange(db.prods.Where(p => p.id == ).Include(p => p.detail)); //用include把关联select出来
db.SaveChanges();
}
Opened connection at // :: +:
SELECT
`Extent1`.`id`,
`Extent1`.`code`,
`Extent1`.`name`,
`Extent2`.`prod_id`,
`Extent2`.`id` AS `id1`,
`Extent2`.`fullDescription`
FROM `prod` AS `Extent1` LEFT OUTER JOIN `prod_detail` AS `Extent2` ON `Extent1`.`id` = `Extent2`.`prod_id`
WHERE = `Extent1`.`id`
-- Executing at // :: +:
-- Completed in ms with result: EFMySqlDataReader Closed connection at // :: +:
Opened connection at // :: +:
Started transaction at // :: +:
DELETE FROM `prod_detail` WHERE `prod_id` =
-- Executing at // :: +:
-- Completed in ms with result: DELETE FROM `prod` WHERE `id` =
-- Executing at // :: +:
-- Completed in ms with result: Committed transaction at // :: +:
Closed connection at // :: +:
Disposed transaction at // :: +:
如果想删除 n-n 关系的rows
Room room = db.rooms.Where(r => r.id == ).Include(r => r.imgs).Include(r => r.facilities).FirstOrDefault();
var facility = room.facilities.First();
//db.Entry(facility).State = EntityState.Deleted; 这样不行哦！
room.facilities.Remove(facility); //关键 : 要这样！
db.SaveChanges();
update 也是先select出来，然后EF会再saveChange时对比之前的值然后生成语句update
using (EFDB db = new EFDB())
{
//db.Configuration.LazyLoadingEnabled = false;
var prod = db.prods.Where(p => p.id == ).FirstOrDefault();
prod.name = "change";
db.SaveChanges();
}
Opened connection at // :: +:
SELECT
`Extent1`.`id`,
`Extent1`.`code`,
`Extent1`.`name`
FROM `prod` AS `Extent1`
WHERE = `Extent1`.`id` LIMIT
-- Executing at // :: +:
-- Completed in ms with result: EFMySqlDataReader Closed connection at // :: +:
Opened connection at // :: +:
Started transaction at // :: +:
UPDATE `prod` SET `name`=@gp1 WHERE `id` =
-- @gp1: 'change' (Type = String, IsNullable = false, Size = )
-- Executing at // :: +:
-- Completed in ms with result: Committed transaction at // :: +:
Closed connection at // :: +:
Disposed transaction at // :: +:
EF 6 默认情况会使用 Transaction 的.
比如当我们对实体做了很多修改，最后调用saveChange().EF会开启Transaction然后执行多个语句，如果其中一个语句有问题，就会rollBack
需要注意的是，每一次saveChange都会开启和关闭一次 tansaction ，所以如果你调用多次的话，自己要搞清楚
using (EFDB db = new EFDB())
{
//db.Configuration.LazyLoadingEnabled = false;
var prod = db.prods.Where(p => p.id == ).Include(p => p.detail).FirstOrDefault();
prod.name = "change22";
db.SaveChanges();
prod.detail.fullDescription = "zzz";
db.SaveChanges();
}
Opened connection at // :: +:
Started transaction at // :: +:
UPDATE `prod` SET `name`=@gp1 WHERE `id` =
-- @gp1: 'change22' (Type = String, IsNullable = false, Size = )
-- Executing at // :: +:
-- Completed in ms with result: Committed transaction at // :: +:
Closed connection at // :: +:
Disposed transaction at // :: +:
Opened connection at // :: +:
Started transaction at // :: +:
UPDATE `prod_detail` SET `fullDescription`=@gp1 WHERE `prod_id` =
-- @gp1: 'zzz' (Type = String, IsNullable = false, Size = )
-- Executing at // :: +:
-- Completed in ms with result: Committed transaction at // :: +:
Closed connection at // :: +:
Disposed transaction at // :: +:
还有一种方式是，让我们自己来控制
using (EFDB db = new EFDB())
{
using (var trans = db.Database.BeginTransaction())
{
try
{
//db.Configuration.LazyLoadingEnabled = false;
var prod = db.prods.Where(p => p.id == ).Include(p => p.detail).FirstOrDefault();
prod.name = "xx1234";
db.SaveChanges();
prod.detail.fullDescription = "";
throw new Exception("x"); //即使这里有问题，上一个saveChange也能rollBack
db.SaveChanges();
trans.Commit();
}
catch (Exception)
{
if(trans.UnderlyingTransaction.Connection != null) trans.Rollback(); //检查确保还没有被 rocllback 或者 commit rollback 2 次是会error的
}
}
}
Opened connection at // :: +:
Started transaction at // :: +:
SELECT
`Extent1`.`id`,
`Extent1`.`code`,
`Extent1`.`name`,
`Extent2`.`prod_id`,
`Extent2`.`id` AS `id1`,
`Extent2`.`fullDescription`
FROM `prod` AS `Extent1` LEFT OUTER JOIN `prod_detail` AS `Extent2` ON `Extent1`.`id` = `Extent2`.`prod_id`
WHERE = `Extent1`.`id` LIMIT
-- Executing at // :: +:
-- Completed in ms with result: EFMySqlDataReader UPDATE `prod` SET `name`=@gp1 WHERE `id` =
-- @gp1: 'xx12234' (Type = String, IsNullable = false, Size = )
-- Executing at // :: +:
-- Completed in ms with result: Rolled back transaction at // :: +:
Disposed transaction at // :: +:
Closed connection at // :: +:
这里补个优化删除，如果只是要删除一条row,而且知道这个row的id，那么就可以这样.
要注意如果sql被改过，比如这id已经被删除，会报错哦。
var address = new Address { addressId = }; //放id
db.addresss.Attach(address); //用Attach
db.addresss.Remove(address); //remove
db.SaveChanges();
删除时我们也可以直接发SQL COMMAND, 因为delete语句比较通俗规范副作用比较少，不然EF不建议我们自己写Command的.
using (DB db = new DB())
{
int x = db.Database.ExecuteSqlCommand("delete from daily_room where id ={0}", );
}
继承类删除 : db.parents.remove(child | parent); 这样parent 和 child 都会一起被删除的.
db.SaveChanges()是通过肮数据监测实现的，一个实体有3个值 : database value, original value, current value
有几种state :
detached = 1 //如果obj并非是under entity的，那么将是detached状态
unchanged = 2 //default
added = 4 //if don havedatabase value
deleted = 8 //if don have current value
modified = 16 //if current value != original base on 不同状态 (下面是我自己的经验，没有参考过源码)
当一个实体是detached时代表它是普通对象(可能是从json转换来的), 这里普通对象的意思是它没有进入 entity 的作用域内 (不要以为一个class有[key],[foreign key])标签 new 出来就是entity对象哦)
我们可以通过 attach 把一个普通对象添加进entity 作用域内
这个attach其实就调用了 db.Entry(p).State = EntityState.Unchanged;
对一个detached设置state时，不同的state是不同的.
set to add , 子层的state也是add (但是如果你先set to unchanged ，在set to add , 子层将会是 unchanged哦)
set to unchanged , 子层也是unchanged
set to modified , 子层是 unchanged
set to deleted , 子层的值会被删除
如果entity set to modified , 它的所有属性的都会set to isModified
如果其中一个attr set to isModified , entity state 会变成 modified
要set attr 的 ismodified ，实体必须在作用域先。
所以通常做法是 :
一个obj set to modified. 子层不用管(除非你要更新等), 哪一些不要update的，set attr ismodified to false
在 attach 时，如果 foreign id 和 子层obj id 不 match ,会报错!
在 saveChanges 时，foreign id 和 子层 obj id 不match 是不会报错的哦.
p.s attach时也是要注意, EF作用域内不可以有相同 Key 的实体哦
如果attach之后，我们必须自己来控制好外键对象和外键值，EF会用跟踪改变来判断最后使用外键值还是外键对象的id来更新数据库，最后在match它们。
obj id 优先于 foreign id 哦
一个错误的例子是 attach之后，我们先修改外键对象去一个(id =5 )的对象，然后我们在修改外键值去 (id = 6), EF会用6作为更新，然后在把外键对象变成 null.
所以不要同时修改2个，很危险.
在做项目的时候，想要替换外键，就不要把外键对象传上来，不然就error了.除非你能一起传新的外键对象。
var obj = new SingleTable
{
data = "",
id =
};
db.SingleTables.Attach(obj); //attach 进entity上下文
db.Entry(obj).State = System.Data.Entity.EntityState.Modified;
db.Entry(obj).Property("data").IsModified = true; //如果只是想修改一些属性罢了的话
db.Entry(obj).Property(o => o.data).IsModified = true;
db.saveChanges();
//顺便提一下 : Entry 还可以控制 database value, original value, current value哦
//db.Entry(obj).CurrentValues.SetValues("x");
//db.Entry(obj).OriginalValues.SetValues("x");
db.Entry(obj).State = System.Data.Entity.EntityState.Modified;
(一次过把所有属性的 isModified set to true (不包括层级对象) , 之后我们也可以单独 set 某些属性回到 false 如果不希望修改)
更新 by detached object
大多时候我们的对象是来自前台的json
一个不再entity context 内的对象，我们称为 detached object (即你用entry会得到state是detached)
using (EFDB db = new EFDB())
{
SingleTable sqlObject = db.SingleTables.Find(detachObject.id); //先从sql获取原本的数据
if (sqlObject == null) //如果没有的话 maybe insert or 不处理
{
db.SingleTables.Add(detachObject);
}
else
{
//透过 entry 设置一个新的value对象进去，那么它在save的时候会dirty check, 如果有modified的它会更新.
var entry = db.Entry(sqlObject); //修改sqlObject的current values
entry.CurrentValues.SetValues(detachObject); //它会dirty. 如果对象中有 array 忽略(skip,当作没看到，完全不处理), object会=null,but complex is ok
}
db.SaveChanges();
}
补充insert:
var employer = db.employers.FirstOrDefault(ee => ee.id == 22);
var company = new Company
{
name = "22",
addressComplex = new addressComplex()
//不需要写 foreign id or obj 没关系
};
employer.companys = new List<Company> //也不需要include出来
{
company
};
await db.mySaveChangesAsync(); //不需要使用db.companys.add, 最后依然可以insert进去sql
可以看得出来，EF的检查还是挺好用的。
补充 update :
//get and update
var FAQSQL = await db.FAQs.FindAsync(id);
db.Entry(FAQSQL).CurrentValues.SetValues(FAQ);
db.Entry(FAQSQL).OriginalValues["rowVersion"] = FAQ.rowVersion; //direct update
FAQ.id = id;
db.Entry(FAQ).State = EntityState.Modified;
get and update 会导致请求多一次，一般情况下是不推荐使用的，但是很多时候由于要兼顾权限管理，我们依然必须使用这个方式，而在使用SetValues的时候rowVersion是不会被更新的，所以我们必须要自己写入
rowVersion 来防止并发。
direct update 就不需要担心并发问题了，但是对权限管理就不行了。
补充 foreign key
如果 foreign obj.id 和 foreign id 不同，saveChange 会用foreign obj.id 然后foreign id 会被刷新掉
更新 2015-11-07:
常用 :
DbContextTransaction trans = null;
trans = DB.Sharing.beginTransaction();
DB.Sharing.commitTransaction(trans);
DB.Sharing.rollBackTansaction();