Add 'metric' argument to Party.decluster()

[cjhopp]

By default, Party.decluster() should now consider the average single-station correlation value. Raw correlation sum can be specified with the metric argument.

[cjhopp]

I think fails are timeouts due to LOOONNNGG subspace tests in time domain. AppVeyor builds passing and subspace tests much faster with eventual pull #92 so this is probably good to go?

[calum-chamberlain]

What are we waiting on for #92? It would be good to merge that before this to make sure that all things pass… From: Chet Hopp <[email protected]> Reply-To: eqcorrscan/EQcorrscan <[email protected]> Date: Thursday, 15 June 2017 at 05:56 To: eqcorrscan/EQcorrscan <[email protected]> Cc: Subscribed <[email protected]> Subject: Re: [eqcorrscan/EQcorrscan] Add 'metric' argument to Party.decluster() (#100) I think fails are timeouts due to LOOONNNGG subspace tests in time domain. AppVeyor builds passing and subspace tests much faster with eventual pull #92<#92> so this is probably good to go? — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub<#100 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AKaGc-GpsIDurC16hBCXe-8DywzVpDCSks5sELlkgaJpZM4N6rXd>.

[cjhopp]

I was just waiting for a review. But I'll merge #92 if you're down.

[calum-chamberlain]

Shall review #92 in the next hour (yay jetlag). From: Chet Hopp <[email protected]> Reply-To: eqcorrscan/EQcorrscan <[email protected]> Date: Thursday, 15 June 2017 at 06:00 To: eqcorrscan/EQcorrscan <[email protected]> Cc: Calum Chamberlain <[email protected]>, Comment <[email protected]> Subject: Re: [eqcorrscan/EQcorrscan] Add 'metric' argument to Party.decluster() (#100) I was just waiting for a review. But I'll merge #92<#92> if you're down. — You are receiving this because you commented. Reply to this email directly, view it on GitHub<#100 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AKaGc_sUvbi2O7hYJf7zmO5oBYWtJHsnks5sELp1gaJpZM4N6rXd>.

[cjhopp]

Hayzeus. I just googled, "What time is it in London?"

.....

[calum-chamberlain]

Looks good to me, but can you add a simple test in to check both options for the metric argument.

[cjhopp]

Surely can! I'll add it at home while watching Peaky Blinders.

[cjhopp]

Aight, there's a shot at it. Not sure if that's an acceptable way to do it? Test are still a bit obscure to me.

[cjhopp]

A real potpourri of failures, but they seem to be ones we've seen before (i.e. weird numpy error for 3.5 on Linux and GeoNet client issues)

[calum-chamberlain]

Travis looks good, but why the appveyor fail? Their mobile site crashes my phone with that long test log, any ideas?

[cjhopp]

================================== FAILURES ===================================
______________ [doctest] eqcorrscan.utils.archive_read.read_data ______________
044 
045     .. note:: A note on arc_types, if arc_type is day_vols, then this will         look for directories labelled in the IRIS DMC conventions of         Yyyyy/Rjjj.01/... where yyyy is the year and jjj is the julian day.         Data within these files directories should be stored as day-long,         single-channel files.  This is not implemented in the fasted way         possible to allow for a more general situation.  If you require more         speed you will need to re-write this.
046 
047     .. rubric:: Example
048 
049     >>> from eqcorrscan.utils.archive_read import read_data
050     >>> from obspy import UTCDateTime
051     >>> t1 = UTCDateTime(2012, 3, 26)
052     >>> stachans = [('FOZ', 'HHZ'), ('JCZ', 'HHZ')]
053     >>> st = read_data('GEONET', 'FDSN', t1, stachans)
UNEXPECTED EXCEPTION: ValueError(u'The current client does not have a station service.',)

[calum-chamberlain]

Just rerun? Calum Chamberlain Postdoctoral Fellow, Seismology Victoria University of Wellington Sent from mobile.

…

________________________________ From: Chet Hopp <[email protected]> Sent: Saturday, June 17, 2017 12:15:19 AM To: eqcorrscan/EQcorrscan Cc: Calum Chamberlain; Comment Subject: Re: [eqcorrscan/EQcorrscan] Add 'metric' argument to Party.decluster() (#100) ================================== FAILURES =================================== ______________ [doctest] eqcorrscan.utils.archive_read.read_data ______________ 044 045 .. note:: A note on arc_types, if arc_type is day_vols, then this will look for directories labelled in the IRIS DMC conventions of Yyyyy/Rjjj.01/... where yyyy is the year and jjj is the julian day. Data within these files directories should be stored as day-long, single-channel files. This is not implemented in the fasted way possible to allow for a more general situation. If you require more speed you will need to re-write this. 046 047 .. rubric:: Example 048 049 >>> from eqcorrscan.utils.archive_read import read_data 050 >>> from obspy import UTCDateTime 051 >>> t1 = UTCDateTime(2012, 3, 26) 052 >>> stachans = [('FOZ', 'HHZ'), ('JCZ', 'HHZ')] 053 >>> st = read_data('GEONET', 'FDSN', t1, stachans) UNEXPECTED EXCEPTION: ValueError(u'The current client does not have a station service.',) — You are receiving this because you commented. Reply to this email directly, view it on GitHub<#100 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/AKaGc2U5WoMp3hBaxEPcpVPx2ujKlxSFks5sEwyHgaJpZM4N6rXd>.

[cjhopp]

Still having permissions issues with AppVeyor. I can rerun builds for successful tests, but not for failed ones. I only see the 'Log' button for failed tests. Looking into it now.

[calum-chamberlain]

That's annoying, I wonder if there is something I need to do.
Anyway, have restarted that build. Should green light, then go for it.

[cjhopp]

Okeedoke. Merging now.

With small changes like this, just a couple of missed lines means missing the codecov target. Am I understanding that right? I did add some additional coverage for test_party_decluster(), but not quite enough.