Ed Hartnett comments

Results 98 comments of


                                            Ed Hartnett

Writing to multiple unlimited dimension variables, seg-fault and/or free() invalid next size

Can we get some C code to reproduce this?

Potential N^2 issue in large file read

Can you try with the code in PR #1262. Can you try with that code? You must recreate you test file with the library after applying 1262. Then you will...

Potential N^2 issue in large file read

Greg, did you recreate your file with the new version of netcdf? Use h5dump to confirm that all datasets in the file have the _Netcdf4Coodinates attr, which lists dimids for...

Potential N^2 issue in large file read

Can you use a profiler and see where all the time is being spent? It should not be in rec_match_dimscales(). So where is it?

Potential N^2 issue in large file read

What do you mean "modify to write the _Netcdf4Coordinates attribute for all variables"? This should be happening automatically.

Potential N^2 issue in large file read

OK, thanks for catching that, it should be used even for 1D vars. I will fix that. Let me construct a test case with 5000 dimensions and see what I...

Potential N^2 issue in large file read

I am modifying the recently-added tst_wrf_reads.c to emulate this case. That way we can try a few things...

NetCDF needs better and faster compression, especially for HPC applications

@dopplershift by "handles unpacking" I mean it will automatically apply the scale/offset parameters and change a packed short into an unpacked float. Am I correct that python netCDF-4 library does...

NetCDF needs better and faster compression, especially for HPC applications

Wow, even more great commentary and information. Clearly this is a topic which generates a lot of interest! @epourmal good suggestions, I will make them issues in the CCR project...

HDF5 stride has different semantics than netcdf stride semantics wrt unlimited.

I would like to fix this.