Thanks to both of you for your suggestions... the context is that a
user of my package (h5py) is retrieving byte streams over a network or
database connection which represent HDF5 files.  It would be nice if
there were a way to open and modify the files in memory, instead of
writing them to the file system, modifying, and then re-reading into
memory.  Since it looks like HDF5 can't do this natively I suppose the
next best thing is to use tmpfs to store the temporary copy of the
file.  Making a new driver is beyond the scope of the project. :)

Andrew

On Thu, Mar 25, 2010 at 1:15 PM, Dimitris Servis <[email protected]> wrote:
> Hi Andrew
>
> you have to play a bit with the driver. I use it to perform HDF5 IO in
> memory with or without file backup. Actually I copy objects from a disk file
> to a memory file (usually I use compression as well), attach the buffer to
> an XML message with some metadata, pass it through the network and read the
> buffer on the other side and write the data into another disk file. Actually
> if there was a function like H5F_set_vfd_handle along with the
> H5F_get_vfd_handle it would work. Unfortunately, as usual, I cannot share
> code....
>
> HTH
>
> -- dimitris

_______________________________________________
Hdf-forum is for HDF software users discussion.
[email protected]
http://mail.hdfgroup.org/mailman/listinfo/hdf-forum_hdfgroup.org

Reply via email to