[SPARK-4553][SQL] Query for parquet table with string fields in spark sql hive get binary result

[scwf]

Link to #3352

[SparkQA]

Test build #23738 has started for PR 3414 at commit 2e04cd3.

• This patch merges cleanly.

[scwf]

fix this

[SparkQA]

Test build #23740 has started for PR 3414 at commit d563124.

• This patch merges cleanly.

[SparkQA]

Test build #23738 has finished for PR 3414 at commit 2e04cd3.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/23738/
Test PASSed.

[SparkQA]

Test build #23740 has finished for PR 3414 at commit d563124.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/23740/
Test PASSed.

[marmbrus]

I don't think we want to set this flag to always be true otherwise its not possible to read binary data anymore. A reasonable thing to do would be override the conf when the convertMetastoreParquet code path is taken so at least then we are faithfully mimicking hive's behavior.

[SparkQA]

Test build #23810 has started for PR 3414 at commit 9c85c22.

• This patch merges cleanly.

[SparkQA]

Test build #23810 has finished for PR 3414 at commit 9c85c22.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[AmplabJenkins]

Test PASSed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/23810/
Test PASSed.

[marmbrus]

We shouldn't be mutating configuration when accessing a setting. Also, I think that #3441 is going to solve this problem so perhaps we can close this issue?

[scwf]

Ok, if @liancheng fix them, i will close this two PR.

[marmbrus]

mind closing this manually? or script seems to be missing it.

[scwf]

ok to close this