[jira] [Comment Edited] (PHOENIX-3396) Valid Multi-byte strings whose total byte size is greater than the max char limit cannot be inserted into VARCHAR fields in the PK
[
https://issues.apache.org/jira/browse/PHOENIX-3396?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15602805#comment-15602805
]
Jan Fernando edited comment on PHOENIX-3396 at 10/24/16 6:25 PM:
-
[~giacomotaylor] Was there a reason this works differently with non-PK and PK
fields? I only see this issue if a field is part of the PK. I think the
behavior should be consistent in all cases.
was (Author: jfernando_sfdc):
[~giacomotaylor] Was there a reason this works differently with non-PK and PK
fields. I only see this issue if a fields is part of the PK. I think the
behavior should be consistent in all cases.
> Valid Multi-byte strings whose total byte size is greater than the max char
> limit cannot be inserted into VARCHAR fields in the PK
> ---
>
> Key: PHOENIX-3396
> URL: https://issues.apache.org/jira/browse/PHOENIX-3396
> Project: Phoenix
> Issue Type: Bug
>Reporter: Jan Fernando
>
> We allow users to insert multi-byte characters into VARCHAR columns that are
> part of a table or view's PK. We noticed that Strings that had a valid number
> of characters (i.e. were less than than max char length) were causing upserts
> to fail with with the following exception:
> Caused by: java.sql.SQLException: ERROR 206 (22003): The data exceeds the max
> capacity for the data type. MYTABLE may not exceed 100 bytes
> ('緓嗝加슪䐤㵞靹疸芬꽣汚佃䘯茵䖻埾巆蕤ⱅ澴粖蟤य褻酃岤豦팑薰鄩脼ժ끦碉ķ窯尬룗㚈Ꝝ퍛爃됰灁ᄠࢥ')
> There appears to be an issue in PTableImpl.newKey() where we check the
> maxLength in chars against the byte length in this check:
> maxLength != null && !type.isArrayType() && byteValue.length > maxLength
> To reproduce you can run the following:
> CREATE TABLE TEXT_FIELD_VALIDATION_PK (TEXT VARCHAR(20), TEXT1 VARCHAR(20)
> CONSTRAINT PK PRIMARY KEY (TEXT));
> UPSERT INTO TEXT_FIELD_VALIDATION_PK VALUES ('澴粖蟤य褻酃岤豦팑薰鄩脼ժ끦碉碉', 'test');
> The string we insert into the column TEXT is 20 chars, but greater than 20
> bytes.
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
[jira] [Comment Edited] (PHOENIX-3396) Valid Multi-byte strings whose total byte size is greater than the max char limit cannot be inserted into VARCHAR fields in the PK
[
https://issues.apache.org/jira/browse/PHOENIX-3396?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15596812#comment-15596812
]
James Taylor edited comment on PHOENIX-3396 at 10/22/16 1:41 AM:
-
Phoenix interpretes the max as bytes instead of characters. That was done on
purpose, but using characters instead would make more sense. As a workaround,
just increase the max length by a factor of 3 - it won't have any impact on
performance. To do that, you can follow the directions outlined here:
http://search-hadoop.com/m/9UY0h21LtMS1KMm942&subj=Re+Can+I+change+a+String+column+s+size+and+preserve+the+data+
was (Author: jamestaylor):
Phoenix interpretes the max as bytes instead of characters. That was done on
purpose, but using characters instead would make more sense.
> Valid Multi-byte strings whose total byte size is greater than the max char
> limit cannot be inserted into VARCHAR fields in the PK
> ---
>
> Key: PHOENIX-3396
> URL: https://issues.apache.org/jira/browse/PHOENIX-3396
> Project: Phoenix
> Issue Type: Bug
>Reporter: Jan Fernando
>
> We allow users to insert multi-byte characters into VARCHAR columns that are
> part of a table or view's PK. We noticed that Strings that had a valid number
> of characters (i.e. were less than than max char length) were causing upserts
> to fail with with the following exception:
> Caused by: java.sql.SQLException: ERROR 206 (22003): The data exceeds the max
> capacity for the data type. MYTABLE may not exceed 100 bytes
> ('緓嗝加슪䐤㵞靹疸芬꽣汚佃䘯茵䖻埾巆蕤ⱅ澴粖蟤य褻酃岤豦팑薰鄩脼ժ끦碉ķ窯尬룗㚈Ꝝ퍛爃됰灁ᄠࢥ')
> There appears to be an issue in PTableImpl.newKey() where we check the
> maxLength in chars against the byte length in this check:
> maxLength != null && !type.isArrayType() && byteValue.length > maxLength
> To reproduce you can run the following:
> CREATE TABLE TEXT_FIELD_VALIDATION_PK (TEXT VARCHAR(20), TEXT1 VARCHAR(20)
> CONSTRAINT PK PRIMARY KEY (TEXT));
> UPSERT INTO TEXT_FIELD_VALIDATION_PK VALUES ('澴粖蟤य褻酃岤豦팑薰鄩脼ժ끦碉碉', 'test');
> The string we insert into the column TEXT is 20 chars, but greater than 20
> bytes.
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
