Locking plan for 6.5
От | Bill Hutto |
---|---|
Тема | Locking plan for 6.5 |
Дата | |
Msg-id | 199902282349.PAA12280@mb3.mailbank.com обсуждение исходный текст |
Список | pgsql-general |
Hi Folks of Concern, I noticed that 6.5 is planned to have row-level locking; and basically allowing all readers and writers to read. Is there going to be any way to implement pessimistic/exclusive locking? I have to deal with many data entry people who have little if any correct notion of what they are actually doing. Point is, when someone starts editing a record manually I want to be able to lock it upon commencement of edits, not upon the actual update. This, to prevent contention on updating, which would not be acceptable. One scenario I have is with Progress (DBMS), in which I can bypass locks (read-only) if I choose. But, for data entry, I need to determine if someone else is currently editing the record that I want to edit. Progress remedies these situations with exclusive locks which can also inform of the userid of the locker. Most importantly, is the inability to BEGIN edits on a record that is already being edited, which prevents all kinds of management problems associated with educating, often times, _temporary_ data entry people about contention problems of this sort. They simply wait, write down information, or in the case of Progress, even call the operator who is locking them out. I certainly don't want them forcing updates or spending time determining whose data is most current, or whether they've changed the same fields, or... It's basically like using a mutex. The user sees the latest version (or WAITS for it), and makes decisions whether to change data based on that. The question would best be put as: Will I be able to prevent others from attempting to manually edit a record that is already being edited? (This, as a built-in feature of the DBMS) Thanks, Bill
В списке pgsql-general по дате отправления: