[netcdf-java] NcML aggregation with custom IOSP


I have a custom IOSP that I am trying to use to aggregate datasets. It works when I use the "location" attribute with local files, but my IOSP can also read data via URLs, bypassing the "location" (which I set to /dev/null) and loading my data (e.g. from an ftp URL) into a Variable cache.

I have a trivial example of trying to use joinExisting with two files with two variables: time and value. In the end, the result successfully aggregates the values, but the times from the first file are repeated for the second half of the aggregation.

My first point of confusion is that open gets called on my IOSP THREE times. (If I don't call "initNetcdfFileCache" it calls open 9 times.) It appears to be random whether the first call reads the first or second file, then the same file gets read again. However, even the working test with "location" also has the third read. That redundant read seems indicative of bigger problems.

Somewhere in the ~20 hops between my read and the IOSP getting invoked (AggregationExisting.buildNetcdfDataset), I see something about a "typical dataset". That seems to be the source of the extra read, and where I gave up so my head wouldn't explode. But I am no closer to understanding the incorrect time values.

Am I barking up the right tree? Should I expect my custom IOSP to work with aggregation?


P.S. I'm using 4.2.

  • 2012 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the netcdf-java archives: