Nothing is ever deleted on the internet! https://web.archive.org/web/20121225230132/http://code.google.com/p/h5py/wiki/FAQ

This made me think, should we arrange a way to get a DOI for h5py? https://guides.github.com/activities/citable-code/. Ping @andrewcollette. I have done this for a couple of my repos.

There's http://joss.theoj.org/ which is more "papery" than Zenodo/Figshare, which may or may not be useful. Depsy is also a thing.

thanks for the links @aragilar, I knew depsy but not joss.

Alternatively

@Book{collette_python_hdf5_2014,
        keywords = {python, hdf5},
        year = {2013},
        publisher = {O'Reilly},
        title = {Python and HDF5},
        author = {Andrew Collette}
}

I used the book citation recently https://arxiv.org/abs/1608.04904 and it seemed a reasonable choice.

Given that h5py hsa its own domain http://www.h5py.org/, that is also a stable citation point.

But @andrewcollette might have an opinion.

For the majority of the projects I am involved with, regardless either there is some proper article (JOSS is IMHO a good venue) we link to zenodo to mint dois for every release. So I would strongly recommend zenodo as a quick and cheap solution until some ultimate citation appears, which would take some time regardless of the chosen publication venue.

Any update here? What to cite? :)

http://dx.doi.org/10.5281/zenodo.6575970 ?

I guess whoever has the authority to decide this is the way to cite h5py could describe it in the docs (most scientific Python packages do this, cfr "citing numpy", "citing scipy", "citing mpi4py" Google results) and close this issue then :]

It's a long time since I contributed, so I don't know who maintains the project and would be able to provide a clear answer. A citing section in the doc would be nice to have indeed.

To contribute something more constructive, I propose possible citation points.

• Andrew Collette's book. Does not reflect current h5py development. Simple to cite.
• The website www.h5py.org . Does not play nice with citations. Requires little work.
• A zenodo "dump" of the code. Datacite citations not as easy as crossref ones. Requires little work.
• A dedicated article (such as JOSS). Requires some work. Simple to cite.

The comments are not a definitive judgement of the different solutions but a starting point for selecting something.

@pdebuyl I opened a pull request for this issue that provides your proposed text in a standard CITATION file. I hope that will at least begin to get this 8 year old issue taken care of.

Hello @erikhuck this is a good idea. As you mention in the PR what is needed is for the maintainers to pick a method. Is there any interest in this @takluyver @tacaswell @aragilar (hoping to have selected appropriately :-) )