Hi Jacob, [Please, try to write to the list from a subscribed address. Thanks!]
> ---------- Missatge transmès ---------- > > Subject: Auto-discard notification > Date: Wednesday 27 May 2009 > From: pytables-users-boun...@lists.sourceforge.net > To: pytables-users-ow...@lists.sourceforge.net > > The attached message has been automatically discarded. > ------------------------------------------------------- > Hi, > > Has anybody had luck attaching large attributes to pytables nodes? I'm > trying to do so but am getting an HDF5 error. The below example crashes > when the dictionary contains ~2550 values or more, but works for fewer. > The message I get is below the code. I know this is not a very performant > way to store large amounts of metadata but it'd be good to know where the > cutoff is between working and crashing. I'm using Mac OS X 10.6, python > 2.5.2, pytables 2.1.1, and HDF5 1.6.7, in case it matters. Yep. The HDF5 1.6.x series has a limit of 16 KB for each attribute [1]. Newer versions of HDF5 may have raised this limit, but I don't know for sure. At any rate, it is very desirable that you use a Leaf instead of an attribute for large datasets. [1] http://www.hdfgroup.org/HDF5/doc1.6/UG/13_Attributes.html Hope that helps, > > Thanks, > Jacob > > ---- > > import numpy as np > import tables as pt > > tbl = pt.openFile('/tmp/test1','w') > node = tbl.createGroup(tbl.root, 'group1') > array = tbl.createArray(tbl.root.group1, 'data', > np.ones((100,100),'float32')) > > testAttr = {} > for i in xrange(10000): > testAttr[i]='value %d'%i > > array.attrs.testAttr = testAttr > > tbl.close() > -------------------------------- > ja...@fudd:~/temp/pyt_attrs$ python test1.py > HDF5-DIAG: Error detected in HDF5 library version: 1.6.7 thread 0. Back > trace follows. > #000: H5A.c line 187 in H5Acreate(): unable to create attribute > major(18): Attribute layer > minor(29): Unable to initialize object > #001: H5A.c line 293 in H5A_create(): unable to update attribute header > messages > major(18): Attribute layer > minor(29): Unable to initialize object > #002: H5O.c line 1275 in H5O_modify(): unable to write object header > message > major(12): Object header layer > minor(25): Write failed > #003: H5O.c line 1394 in H5O_modify_real(): unable to create new message > major(12): Object header layer > minor(29): Unable to initialize object > #004: H5O.c line 1658 in H5O_new_mesg(): object header message is too > large > major(12): Object header layer > minor(29): Unable to initialize object -- Francesc Alted ------------------------------------------------------------------------------ Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers & brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, & iPhoneDevCamp as they present alongside digital heavyweights like Barbarian Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com _______________________________________________ Pytables-users mailing list Pytables-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/pytables-users
