Open lbesnard opened 8 years ago
Thanks for the bug report on this; we will look into it and follow up as soon as we can (I've been traveling for a conference all week). As our resident DAP developer/expert, I wonder if @dmh has any insight into this.
I ran your test, and the results were identical using the latest netcdf-c on github.
G'Day
I may have found an error running a CF compliance checker (from IOOS) on a gridded file. I noticed that some
byte
types are "modified" toshort
if the netcdf file is being accessed via opendap rather than a filesystem/fileserver.See this little bash script which shows the problem for the same file being accessed by the two different protocols : https://gist.github.com/lbesnard/3444799f58e7d75d18a4/revisions
Follows is the vimdiff output of the given script above. See line 106 difference ( -127s vs -127b) as well as line 16 :
I also tried doing a similar thing using the netCDF4 python package, exactly the same thing.
Cheers