Upgrade Parquet to version 1.12.2

[shangxinli]

No description provided.

[shangxinli]

@aweisberg, can we build a staging jar to have a try in presto repo, like we did when upgrading to 1.11.x? Thanks in advance!

[aweisberg]

I deployed hive-apache-3.0.0-8-20220307.153144-32

Thanks!

[shangxinli]

Thanks @aweisberg! I created PR with this version but found building errors because of the zstd-jni duplicate but version mismatch. So I just add a fix here to exclude zstd-jni in parquet. Can you help to release another new version to try out?

[aweisberg]

Deployed 3.0.0-8-20220309.171836-34

[shangxinli]

Thanks. This new version succeeded (PR). Can you merge and release an official version?

[aweisberg]

OK, LGTM. Will release.

[aweisberg]

I just released 3.0.0-8, it should appear in mirrors soon.

[7c00]

@shangxinli Do we really need to upgrade Avro? After upgrading avro to 1.10.1, Presto fails to query Hudi MOR tables. The reason is that there is a breaking change in GenericData class

Old version:

    @Override
    public Object get(String key) {
      Field field = schema.getField(key);
      if (field == null)
        return null;
      return values[field.pos()];
    }

New version:

    @Override
    public Object get(String key) {
      Field field = schema.getField(key);
      if (field == null) {
        throw new AvroRuntimeException("Not a valid schema field: " + key);
      }
      return values[field.pos()];
    }

ref:

prestodb/presto#17463
https://github.com/prestodb/presto/runs/5581838178?check_suite_focus=true

CC: @codope @pratyakshsharma

[shangxinli]

In that case, we can upgrade avro in Hudi to support that? I guess we need to upgrade sooner or later.