[Python] Support read_pandas on a directory of Parquet files

[asfimport]

Reporter: Phillip Cloud / @cpcloud
Assignee: Wes McKinney / @wesm

_{Note: This issue was originally created as ARROW-1041. Please see the migration documentation for further details.}

[asfimport]

Wes McKinney / @wesm:
@cpcloud can you update the scope of this issue?

[asfimport]

Phillip Cloud / @cpcloud:
I think the scope here is pretty limited. read_pandas should support reading a directory of parquet files, which isn't implemented. I can take a look at implementing this at the beginning of next week. It looks like the work involves some forwarding of flags plus making any changes to read_row_group that would prevent us from reading index columns into a pyarrow.Table.

[asfimport]

Wes McKinney / @wesm:
Cool; I may get to it this weekend depending on how the rest of the 0.5.0 issues go, otherwise I'll wait for you

[asfimport]

Wes McKinney / @wesm:
I'm touching this code a bunch working on ARROW-1103 so let me take a crack at this

[asfimport]

Wes McKinney / @wesm:
Included in PR: #862

[asfimport]

Wes McKinney / @wesm:
Resolved in 362e754