| |
| |
| |
|
Page: 1 2 3 4 5
Comments:
<0> if you want a sequential key then you need to either rebuild after each delete, fill in the gaps when you insert, or not delete at all
<0> I suggest you find a better method to do what you want
<0> which sounds like paging to me
<1> yea :)
<2> yes, i know, but this is a readonly database. No delete will be done
<0> so why are there gaps in the first place then?
<2> because of failing insert
<2> sometimes inserts violates uk constraint
<3> how often do you plan to reorder ?
<2> and i lose the sequential
<2> I plan a reorder once a month
<4> The pk means nothing. I'd suggest limiting based on a field with actual meaning.
<0> do you have any foreign keys that refer to that key?
<0> in other tables
<2> yes, one FK but it is setted as cascade
<3> you can add another field in your table if you only plan to reorder once a month
<3> then use that new field as your order
<3> schedule a job to rewrite the order
<2> But there are not a built in function in sql 2005 to rebuild an identity column?
<0> no
<2> :(
<2> it is not a big problem but I'm near the int value limit
<0> you got 2 billion records in the database (or failed inserts)
<2> yes, 1.5 billions
<0> then perhaps you shouldn't have used "int" as your pk then?
<2> the problem is that there are only 200 millions records, but I have got a lot of failed inserts and the pk is actually 1.5 billion
<0> why does so many inserts fail?
<2> because customers did some errors importing files.
<2> and they imported the same files 3 or more times
<0> perhaps you should beef up the validation in those imports then?
<2> yes, LVK, you are right. But the mistake is already done
<2> :(
<0> well, the "reorder ploy" is the worst you can do
<0> I would suggest you change to bigint for the pk
<0> and fix the validation in those import routines
<2> ok LVK.
<2> I will insert a select before the insert
<2> if the count is 1 then no insert will be done
<2> ok?
<5> Do you think there will be a time that we can store serialized objects in a db
<2> LVK, performance will vary using bigint?
<2> I have another questions. The full text search function of sql 2005 is fast than the like keyword?
<0> faster, yes
<6> w00t
<6> people owned
<2> LVK, there was a net split before and I can't say you thank you. Thanks for your help.
Return to
#sql
or
Go to some related
logs:
how to update sequnce next value in oracle
#politics
#politics
#computers
#politics
#windowsxp
#beginner
#freebsd
abraham kill me your son
#microsoft
|
|