I’ve created a rather large hdf5 file, using the h5py python package. It contains one data set with a compound data type with ~120 fields. I try to use HDFView 3.1.4 to view the data. I can open the file, but when I double click on the data set, I get the error message " *** Error occurred while instantiating TableView class ***".
Is this a bug in HDFView? How could I get more details about what went wrong?
If the dataset is really large and/or has to do a lot of translation od data, hdfview will run into memory problems.
Recently fixed/WIP has fixed some issues to do with VLEN types in container types.
Another option (depending on your platform) is to enable the logging, see the logging document in docs folder (or file included within binary) of source.
Well, it’s 2 GB large. One data set containing 2097152 compound objects with ~256 members. For the purpose of HDF5, I would think that this is still small, right?
h5dump and h5debug are not complaining about the file. Looking with h5dump into the file, I find the expected data.
Is there an other HDF5 viewer available, that can be recommended?
I will try to enable the logging (most likely next week).
hdfview (contrary to h5dump) keeps everything in memory, so large files/object counts will cause issues depending on the machine. Lack of resources to convert the object handling processing to a “paging” technique has been the problem. hdffview does more then just show the data, and years ago, it wasn’t a problem.
Hi,
I did a simple test and changed my tool to generate a file that is just a 16th of the original. That file can then be opened and the data can be viewed.
@torsten.robitzki, thank you so much for uploading the large file! I to test big file. Our network is slow 400KB/sec so I’m still downloading it. Downloading will take another hour.
I’ll post test result as soon as downloading finishes. Please stay tuned!
Thanks @christian.oyarzun for the tip!
Yes, it loaded after 2 minutes.
Plotting a line from table … I don’t know when it will end.
Patience is a virtue.