From: Francesc A. <fa...@ca...> - 2007-10-17 19:07:12
|
A Wednesday 17 October 2007, Darren Dale escrigu=E9: > > Mmm, I'm curious about which datatype should be causing this. Can > > you try applying the attached patch and send me the output? > [snip] > datatype--> f > datatype--> <---- > Edatatype--> f <---- Well, this seems to correspond to a complex type. Apparently the go and=20 back trip to/from a netcdf3 file for complex types doesn't work as it=20 should on your machine. The fact is that the netcdf3 module is becoming more and more dangerous=20 as the Numeric package is ageing. Perhaps we should start to think in=20 deprecate it (unless the original contributor, Jeff Whitaker, would=20 like to contribute a complete replacement of Numeric by NumPy for it). By the way, if what you want is to access netcdf3 files, maybe you might=20 want to look at the netcdf4-python project=20 (http://code.google.com/p/netcdf4-python/) which is being led by the=20 very same Jeff that contributed the netcdf3 module for PyTables. > > > Is the netcdf error a known pytables bug, or is it maybe a > > > configuration issue on my end? > > > > I don't know. The NetCDF module still relies on having a proper > > installation of Numeric, and I know that Numeric has issues in > > 64-bit machines (specially when using Python 2.5). > > I know, I'm the one who discovered the issue. The folks at gentoo > created a patch for Numeric, so it will run and pass all the tests on > a 64 bit machine with python-2.5. I guess it is still possible that > there are some issues not addressed by the patch and not exposed by > Numeric's tests. Yes, I'm afraid that there might remain non solved issues with Numeric=20 with 64-bit platforms. > The manual does not indicate that Numeric is required for netcdf3. You are right. I've added a ticket on this:=20 http://www.pytables.org/trac/ticket/124 Cheers, =2D-=20 >0,0< Francesc Altet =A0 =A0 http://www.carabos.com/ V V C=E1rabos Coop. V. =A0=A0Enjoy Data "-" |