added a compression argument to to_csv to be sent to _get_handle

[yoavram]

Writing csv files with gzip compression is very useful as it both minimized the disk size taken by large data files, and can be easily read by R, often faster than the same csv file without compression. 
Reading compressed csv files in R is done without any additional input by the user, and is also already supported in Pandas

[wesm]

Needs a test case

[yoavram]

Can you suggest where a test case should be written?

[ghost]

pandas/tests/test_frame.py has all the test_to_csv_* tests.

[ghost]

I intended to write a roundtrip test, but from_csv doesn't support compression
despite it's hiding down the call stack.
Should probably keep things consistent and update from_csv the same way.

[yoavram]

OK. This is on my task list, but unfortunately not high enough...
Hopefully I will get to it soon.

[ghost]

MIA.

[yoavram]

@y-p I'd like to reopen this and complete the PR.
I notice that read_csv quietly reads gzipped csv files, so if I understand correctly, I need to add a test for both to_csv using a roundtrip with read_csv in pandas/tests/test_frame.py.
Let me know if this is true and if this is still relevant.
Also - should I do anything because so much time passed, like rebase or just fork again and apply my minor changes, or would this be handled by you during merging?

[jreback]

you should rebase on master and open a new PR
quite a lot has changed since this issue came up

contributing docs are here

and this issue is #7615