report the error as well as the line that caused it

[abridgett]

I don't know Scala (at all!) so there's almost certainly cleaner ways - my apologies. The logging at the moment is sometimes unhelpful as it's hard to see the real issue - with DROPMALFORMED you see the line, with another parsing mode you get the error but not the line.

[HyukjinKwon]

Please check travis. [error] /home/travis/build/databricks/spark-csv/src/main/scala/com/databricks/spark/csv/CsvRelation.scala:176: File line length exceeds 100 characters

[HyukjinKwon]

For me I think this is reasonable because I sometimes see some users have some difficulties to find out why it was dropped or malformed.

[codecov-io]

Current coverage is 85.86%

Merging #258 into master will decrease coverage by -0.21% as of 26a1447

@@            master    #258   diff @@
======================================
  Files           12      12       
  Stmts          517     552    +35
  Branches       149     160    +11
  Methods          0       0       
======================================
+ Hit            445     474    +29
  Partial          0       0       
- Missed          72      78     +6

Review entire Coverage Diff as of 26a1447

Powered by Codecov. Updated on successful CI builds.

[abridgett]

Thanks Hyukjin, I've fixed that overlong line

[HyukjinKwon]

Personal opinion: since this is an warning message, shouldn't we maybe just print out the message only by e.getMessage() excluding the class name of the exception?

[HyukjinKwon]

In addition, you can write s"Number format exception (${a.getMessage})." instead of concatenating them with + which is not used generally in this library and Spark codes as well.

[falaki]

This line exceeds 100 characters as well.