[
https://issues.apache.org/jira/browse/HIVE-867?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12767678#action_12767678
]
Zheng Shao commented on HIVE-867:
---------------------------------
If you almost always need a String parameter, we can just use "String" as the
type of the parameter in the UDF definition.
If you almost always need to return a String, we can also just return "String".
So for UDFLeft and UDFRight, we can do:
{code}
public String evaluate(String s, IntWritable r);
{code}
instead of
{code}
public Text evaluate(Text s, IntWritable r);
{code}
This will save a lot of conversions if user do "left(right(col, 10), 3)".
This is the same for the SerDe - for example, RegexSerDe returns "String"
instead of "Text", so "left(col, 3)" where col is from a RegexSerDe table does
not need a conversion from "String" -> "Text" to pass to the Left function, and
then "Text" -> "String" inside the left function.
Of course, the most efficient way is to do the char counting without UTF-8
encoding/decoding, (then we still prefer Text because we don't need to create
new objects), but I think we can do that later unless you want to do it now.
> Add add UDFs found in mysq
> --------------------------
>
> Key: HIVE-867
> URL: https://issues.apache.org/jira/browse/HIVE-867
> Project: Hadoop Hive
> Issue Type: New Feature
> Reporter: Edward Capriolo
> Assignee: Edward Capriolo
> Attachments: hive-867-1.diff, hive-867-2.diff, hive-867-3.diff,
> hive-867-7.diff
>
>
> Some UDF's that mysql has that hive does not.
> atan
> aes_decrypt
> aes_encrypt
> bit_and
> bit_count
> bit_length
> bit_or
> bit_xor
> char_length
> char
> character_length
> collation
> compress
> crc32
> encode
> encrypt
> format
> greatest
> in
> inet_oton
> inet_ntoa
> match
> md5
> oct
> ord
> pi
> radians
> sha1 _sha
> sign
> sleep
> truncate
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
