[Pytables-users] Re: Matlab experience? (Achim Gaedke)

[Dav C. <da...@al...>]

I'm writing a client right now in MATLAB that communicates with a  
repository system written in python.

Generally, MATLAB support is more robust, but far less intelligent  
than PyTables.  For example, individual arrays of a VLArray are  
loaded by matlab as a structure array with a separate name and data  
chunk for _each_ atom.  So, you need to cat them all together  
yourself.  I'm in touch with Mathworks about how clunky this stuff  
is.  If you have a support contract - I'd suggest bugging them.  This  
kind of thing actually matters with them.

I don't know if this is useful, but you might refer to this: Service  
Request # 1-2JECTC regarding 'working with imported HDF5 data'.

Generally, if you keep everything simple - strings, basic arrays and  
scalars, matlab and pytables both read things fine.  However, you  
might want to read the PyTables manual on how to properly set  
metadata for things like hdf5 compounds and such, so that pytables  
will read them.  PyTables default behavior with things like compounds  
is not very robust at this point (so far as I can tell - it won't  
read them without proper metadata).

Cheers,
Dav Clark
917-544-8408