![]() | |
#1
| |||
| |||
|
|
It's a great deal but it also is a feature that needs different logic when implementing an application since you might not get the results you would get by using a 'traditional' locking model. |
#2
| |||
| |||
|
|
nikosv wrote: It's a great deal but it also is a feature that needs different logic when implementing an application since you might not get the results you would get by using a 'traditional' locking model. I suppose that might be a problem if you are porting an application that has depended on some otherwise undesirable quirk of locking, but I can't see it as a valid criticism of MVCC. (But as I say, it looks like Ingres is going to give us the option to lock or to use MVCC.) -- Roy |
#3
| |||
| |||
|
|
On Jul 16, 2009, at 5:03 PM, Roy Hann wrote: nikosv wrote: It's a great deal but it also is a feature that needs different logic when implementing an application since you might not get the results you would get by using a 'traditional' locking model. I suppose that might be a problem if you are porting an application that has depended on some otherwise undesirable quirk of locking, but I can't see it as a valid criticism of MVCC. (But as I say, it looks like Ingres is going to give us the option to lock or to use MVCC.) -- Roy Both of you are assuming the average programmer can spell the word "concurrency". |
|
Lastly, there is some hope. The new web frameworks like Grails include this nice O/R mapping tool layered on Hibernate that encourages the use of a "version" column and uses it for optimistic locking. On top of that, Grails actually has a transaction manager and treats a very complex operation as a ... *gasp* ... TRANSACTION. Imagine that. AND Grails/GORM encourages FOREIGN KEYS, unlike the nasty O/R mapping in Ruby. |
|
Pretty soon, data modeling will be back in vogue! Erwin sales will skyrocket! |
![]() |
| Thread Tools | |
| Display Modes | |
| |