Feature request for h5copy: clobber mode

Hdf-forum archives
1

Would you please consider adding an "overwrite" or "clobber" flag to h5copy?

Unless I've missed it, there's no way to overwrite a dataset with
another unless the target is first removed. And, unless I've missed
it, there's no out-of-the-box "h5rm" utility for accomplishing that
forcing users to each roll their own (e.g. [1]). The use case I have
in mind is transferring simulation settings or initial conditions from
one file into another.

Thanks,
Rhys

[1] https://github.com/RhysU/suzerain/blob/master/postproc/remove_hdf5.py

2

Hi Rhys,

I added both your requests for h5dump and h5copy to JIRA (HDFFV-8759 and HDFFV-8760). No promises, but your request is timely since we will be planning our next release in two weeks.

Thank you!

Elena

···

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Elena Pourmal The HDF Group http://hdfgroup.org
1800 So. Oak St., Suite 203, Champaign IL 61820
217.531.6112
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

On Apr 28, 2014, at 10:58 AM, Rhys Ulerich <rhys.ulerich@gmail.com> wrote:

Would you please consider adding an "overwrite" or "clobber" flag to h5copy?

Unless I've missed it, there's no way to overwrite a dataset with
another unless the target is first removed. And, unless I've missed
it, there's no out-of-the-box "h5rm" utility for accomplishing that
forcing users to each roll their own (e.g. [1]). The use case I have
in mind is transferring simulation settings or initial conditions from
one file into another.

Thanks,
Rhys

[1] https://github.com/RhysU/suzerain/blob/master/postproc/remove_hdf5.py

3

Hi Rhys,

I would like to ask for a clarification:
Do you mean to overwrite just the data of an existing dataset?
or
Do you mean to change even its metadata? E.g., dataset1 is a {10, 10} array of integers
and you want to change it to a {128, 128} array of floating points.

-Albert Cheng

···

On Apr 28, 2014, at 10:58 AM, Rhys Ulerich <rhys.ulerich@gmail.com> wrote:

Would you please consider adding an "overwrite" or "clobber" flag to h5copy?

Unless I've missed it, there's no way to overwrite a dataset with
another unless the target is first removed. And, unless I've missed
it, there's no out-of-the-box "h5rm" utility for accomplishing that
forcing users to each roll their own (e.g. [1]). The use case I have
in mind is transferring simulation settings or initial conditions from
one file into another.

Thanks,
Rhys

[1] https://github.com/RhysU/suzerain/blob/master/postproc/remove_hdf5.py

_______________________________________________
Hdf-forum is for HDF software users discussion.
Hdf-forum@lists.hdfgroup.org
http://mail.lists.hdfgroup.org/mailman/listinfo/hdf-forum_lists.hdfgroup.org

4

Do you mean to change even its metadata?

I was thinking something like the semantics of 'rm -rf' followed by 'cp
-a'. So, a recursive delete and then metadata preserving copy.

That said, I haven't given the metadata angle much thought. What's your
preference?

- Rhys

5

So, a recursive delete and then metadata preserving copy.

Possibly another way to arrive at that conclusion is that running
h5copy --clobber should cause h5diff to report no differences between
the source and the target.

- Rhys