Add support for NCLOB column type #563
Conversation
This uses the `TO_CLOB()` Oracle function to convert NCLOB values into CLOB values. Doc: https://docs.oracle.com/database/121/SQLRF/functions218.htm
|
Thanks for the initiative! |
intgr
force-pushed
the
add-nclob-type-support
branch
from
0e1bdd4 to
2a88233
Compare
|
Added tests. Inserting seems to work fine using the same logic as CLOB type. Or are there any gotchas that didn't occur to me? |
intgr
force-pushed
the
add-nclob-type-support
branch
from
d19047d to
8bd1a2c
Compare
oracle_fdw.c
Outdated
| @@ -2855,6 +2856,8 @@ char | |||
| else if (fdwState->oraTable->cols[i]->oratype == ORA_TYPE_TIMESTAMPLTZ) | |||
| /* convert TIMESTAMP WITH LOCAL TIME ZONE to TIMESTAMP WITH TIME ZONE */ | |||
| format = "%s(%s%s AT TIME ZONE sessiontimezone)"; | |||
| else if (fdwState->oraTable->cols[i]->oratype == ORA_TYPE_NCLOB) | |||
| format = "%s(TO_CLOB(%s%s))"; | |||
There was a problem hiding this comment.
Seems like the parenthesis around TO_CLOB() aren't really needed, I will get rid of them.
|
Thanks! I'll take a close look at that when I come back from vacation on Jan 9. |
There was a problem hiding this comment.
Your patch looks fine.
I am somewhat surprised that data modifications work without any further ado.
Since you would like me to pull your change request, would you accommodate me in these things:
-
please provide a single commit with all your changes, rebased to current git HEAD
-
please update the documentation: remove the note that NCLOB is not supported and add a line to the conversion table
If you feel like it, you can also create a CHANGELOG entry. Otherwise, I will add that in a separate commit.
| @@ -5057,7 +5063,8 @@ checkDataType(oraType oratype, int scale, Oid pgtype, const char *tablename, con | |||
|
|
|||
| /* VARCHAR2 and CLOB can be converted to json */ | |||
|
Are you going to send an updated patch? |
This uses the
TO_CLOB()Oracle function to convert NCLOB values into CLOB values.Doc: https://docs.oracle.com/database/121/SQLRF/functions218.htm
Do you see any downsides to doing this?
Only tested with SELECT & VACUUM.
The comment in the NCLOB handling code previously stated:
I have not verified whether this trick solves the character set conversion issue.