Calum Chamberlain

> No idea why that `.type` is there in the first place.. comparing `.dtype == np....` would be what I'd do Me too. I will have a play today and...

So on Windows the data that come in have: ```python trace.data.dtype == np.int32 trace.data.dtype.type == np.intc ``` On Linux the same script gets data of the form: ```python trace.data.dtype ==...

Just a +1 from me (sorry for no help on this yet): I have run into seg-faults when reading mseed files using a threadpool (without an SDS client) as well...

I like this for the pros, particularly shielding people from the inner workings of objects. However, is calling external things (such as hypoinverse) something that you see coming under obspy...

It would certainly be useful for EQcorrscan - but at the moment things are handled by errors within EQcorrscan (although those error messages could be better). Are you proposing that...

> Exactly. Obspy would use the validator logic internally and why not expose it to make it easier for obspy dependent packages to use as well? This should be fine...

This didn't end up being tackled in #1924, but it would still be good to do at some point.

Can you provide code to replicate this error please so that we can try to find the source of this issue.

Thanks for that @megies, I will have a look there and crack on with trying to speed it up.

Thanks for those comments guys - I thought I could get around the GIL with more direct access to lxml (which I think releases the GIL?), but it would require...