Uh oh!
There was an error while loading. Please reload this page.
parsing options and serializing arrays - #113
Conversation
falaki
commented
Jul 30, 2015
Please update it to rebase against master for tests to run. Also please revert changes to |
this works everywhere but travis! the partitions are re-ordered somehow methinks. leaving this for another time
codecov-io
commented
Aug 1, 2015
Current coverage is |
mohitjaggi
commented
Aug 5, 2015
@falaki should be ready to merge now. |
falaki
commented
Aug 5, 2015
@mohitjaggi This is fairly large. I am about to publish a release with schema inference and all the recent improvements, and then I will review this. |
falaki
commented
Aug 13, 2015
@mohitjaggi this is packing too much into a single PR. Would you please split it. Please first submit one for number parsing improvements and another for arrays. On arrays within CSV it would be good to post an issue and gather some feedback from community first. |
mohitjaggi
commented
Aug 13, 2015
will do On Thu, Aug 13, 2015 at 10:14 AM, Hossein Falaki notifications@github.com
|
mohitjaggi
commented
Aug 13, 2015
see #124 |
several parsing options are added. they are organized in classes because there are many of them. a "text" based API to configure options is provided.
another feature is the ability to serialize a column of arrays. the array is "unnested" and column names to use are supplied by user. this is useful for writing out csv after doing transforms on the data that "expand" the number of columns e.g. one hot encode a category. this can be improved later. e.g. sparse vector from mllib can replace the array.