You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@avro.apache.org by "Ryan Skraba (Jira)" <ji...@apache.org> on 2023/06/16 19:07:00 UTC

[jira] [Created] (AVRO-3783) [Java] Deserialize byte lengths as LONG but limit to INT

Ryan Skraba created AVRO-3783:
---------------------------------

             Summary: [Java] Deserialize byte lengths as LONG but limit to INT
                 Key: AVRO-3783
                 URL: https://issues.apache.org/jira/browse/AVRO-3783
             Project: Apache Avro
          Issue Type: Bug
            Reporter: Jack Klamer
            Assignee: Jack Klamer
             Fix For: 1.11.2


In the spec, the encoding for the bytes type is bytes are encoded as a long followed by that many bytes of data.. In the Java binary decoders they are read as ints this is not a correctness issue because the VLE of the long is the same as that of an equivalently valued int. The int is used (I assume) to enable easier interop with the ByteBuffer java class. But in the rare cases where validly encoded data of more than MAX_ARRAY_SIZE bytes is found, this change will cause an error of

{code}
throw new UnsupportedOperationException(
          "Cannot read arrays longer than " + MAX_ARRAY_SIZE + " bytes in Java library");
{code}

instead of

{code}
throw new InvalidNumberEncodingException("Invalid int encoding");
{code}

This type of checking is consistent with what happens in readString which reads its size as a long.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)