Lion Krischer

I'm happy to help with py2/3 compatibility. Dummy question: What is the difference between the datamark and win32 format? I thought they are the same to some extent? CC @ThomasLecocq

Should we also rename `io.datamark` to `io.win` and maybe somehow merge the win and win32 data format if they share a lot of characteristics. Would be easier to maintain in...

Sorry -- accidentially closed it by pressing the wrong button.

We should also add this to the non-routed clients. Could be some kind of mix-in that we can just tag on to all the classes.

Also lets move this to 1.2.0 - this i pretty subtle to do correct and we really have to get 1.1.0 out.

There are a bunch of `multidict` implementations out there which would solve this in a fully backwards compatible manner. We could just roll our own if we don't want another...

:+1: for using lists! Definitely has less complexity!

Hmm...no lets leave this open. I did not look at the `resample()` method for a long time (I'm just using lanczos nowadays) but the issue with the frequency domain interpolation...

Yea I could also never get it to work across all Python version on OSX. I know pytest can capture everything so it must be possible. There is also this...

It could also be done with the `CoefficientsTypeResponseStage`: http://docs.obspy.org/packages/autogen/obspy.core.inventory.response.CoefficientsTypeResponseStage.html The SEED manual does not really recommend that for numerical stability reasons but I don't really see why this should matter...