On 07/12/2017 09:31 PM, Tom Lane wrote:
rihad <rihad@xxxxxxx> writes:
On 07/12/2017 01:54 PM, Albe Laurenz wrote:
As you see, your index is still sorted according to the C collation
and scanning it returns wrong results.
This ordering issue can certainly be classified as an inconsistency, but
nothing to lose sleep over. Is this all that is normally meant when
saying "index corruption"?
Laurenz neglected to point out that if the index isn't sorted the way that
the system assumes it is, then searches may fail to find values that are
present (due to descending into the wrong subtree), and by the same token
insertions may fail to enforce uniqueness. That's pretty corrupt in
my book.
regards, tom lane
What if only English letters are used in the textual indices (ascii
0-127), would they still be impacted after datctype&datcollate
"C"->"en_US.UTF-8" change? Encoding has always been UTF8, btw.
postgres=# \l
List of databases
Name | Owner | Encoding | Collate | Ctype | Access
privileges
-----------+----------+----------+-------------+-------------+-----------------------
mydb | myuser | UTF8 | C | C |
--
Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-general
