[
https://issues.apache.org/jira/browse/DERBY-3741?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12615304#action_12615304
]
Kathey Marsden commented on DERBY-3741:
---------------------------------------
Suran asked me off-line for some pointers on this issue, but I am posting my
response to the issue to keep the discussion on the list and allow room for
anyone to correct me if what I say is wrong.
Probably the first best approach is to write the test. Write a test ClobMemTest
which is similar to BlobMemTest and add it to the memory suite. Make sure you
have some multibyte characters so we are sure we are getting the character
length and not the byte length when we fix this. The test should pass normally
but fail when run with the target junit-lowmem.
As for the code change, I think we are looking to change SQLChar.getLength(). I
believe the character length is encoded in the stream in some cases so we want
to retrieve that if possible and if not read (skip) the entire stream to get
the character length. Either way we will need to reset the stream in the end to
make sure we are positioned back at the beginning of the stream. The changes
should be the same as those made to SQLBinary.getLength() for BLOBS
(DERBY-3732) but not exactly the same because we are dealing with characters.
Hope this helps. Please let us know as you have more questions.
> SQL LENGTH function materializes CLOB into memory
> -------------------------------------------------
>
> Key: DERBY-3741
> URL: https://issues.apache.org/jira/browse/DERBY-3741
> Project: Derby
> Issue Type: Bug
> Components: SQL
> Affects Versions: 10.3.3.0, 10.4.1.3, 10.5.0.0
> Reporter: Kathey Marsden
> Assignee: Suran Jayathilaka
> Attachments: LargeLengthClob.zip
>
>
> Similar to DERBY-3732, the SQL LENGTH function also materializes CLOB's into
> memory. See attached repro.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
