Closed jasonsurratt closed 8 years ago
Maybe test an rgb 3-band and a single-band image, if that's not overkill.
Perhaps a test against 16-bit, 11-bit (signed/unsigned), 8-bit data . Most rgb 3-band will be 8-bit I assume as it's not for analysis, just display so in the end that might be overkill.
Do we have any test images anywhere?
@mattjdnv I have just the single test image (3-band WV-2 rgb of Rome) that @jchoi1267 provided when we first implemented the functionality. I suppose we can track down more but I'd like to see what the bounds for the range of imagery datasets supported will be first before searching for data.
I think starting with that should be fine. Just crop it down and add it. This should find gross problems like missing gdal/python bindings. As new specific requirements pop up we can add more image types.
It's cropped already to the colosseum and pretty small (11MB). I can make it smaller if needed. Currently it is in gis-data/eucom/italy
The smallest thing that can possible create multiple tiles would be best. I'm allergic to repo bloat. <achew!>
ok will reduce it further...
@mattjdnv RomanColosseum file is reduced to 475KiB. Tested in the UI and is producing tiles in hoot/ingest/processed/BASEMAP/filename..
Excellent! Is it on Selma?
@mattjdnv same path as shown above [gis-data/eucom/italy]
@mjeffe98 Got it. Thanks.
Merged
@mattjdnv Could you give this a whack? I've noticed a few tickets floating around with CentOS not being configured properly for gdal2tiles. Adding a cmd test for this should be rather straight forward and will flag a bad CentOS install when the tests are run. Keep it small and fast if you can.
@jchoi1267, @brianhatchl, @mjeffe98 Do you have any specifics you would want in this test?