aodn / aodn-portal

AODN Open Geospatial Portal
https://portal.aodn.org.au/
GNU General Public License v3.0
21 stars 13 forks source link

GoGoDuck download fails with incorrect time values #2052

Closed rogerproctor closed 8 years ago

rogerproctor commented 8 years ago

Selecting a single time for start finish returns an error ... see log below

The list of URLs obtained was empty, were your subseting parameters OK?
The collection name was 'srs_sst_l3s_1d_day_gridded_url'
The subset parameters were 'TIME,2015-11-01T03:20:00.000Z,2015-11-01T03:20:00.000999Z;LATITUDE,-90.0,90.0;LONGITUDE,-180.0,180.0'
dnahodil commented 8 years ago

This is certainly an issue (and I was able to reproduce the same behaviour). You can see the problem in the subset parameters:

... TIME,2015-11-01T03:20:00.000Z,2015-11-01T03:20:00.000999Z; ...
                                                      ^^^^^^

We'll need to track down where that error is coming from. We munge dates by adding 999 milliseconds in GogoduckService to account for date formatting errors (https://github.com/aodn/aodn-portal/blob/325a5e0285bf0debbc209952e8d1e0e734100e15/grails-app/services/au/org/emii/portal/GogoduckService.groovy#L43). Perhaps those date formats have changed in the meantime...

dnahodil commented 8 years ago

See https://github.com/aodn/go-go-duck/issues/113 regarding the size discrepancy.

@danfruehauf I've removed some of our comments which are no longer related to this bug.

dnahodil commented 8 years ago

Turns out that the crazy date munging isn't actually causing the issue. Although we will need to fix that anyway as it is incorrect and might be causing bugs we don't know about.

dnahodil commented 8 years ago

Given that we need to support a few flavours of ncWMS we probably need to keep this munging code in, but ensure that it works for both formats we expect (with and without the milliseconds component).

dnahodil commented 8 years ago

Ok, the problem is actually a content problem (see https://github.com/aodn/content/issues/149). Loz is going to look into that. Also, it seems that the munging code (as it is currently) is actually still fine so I won't make any changes there.