HDF

[Andy Adler]

I wanted just to make a quick comment. I'm at the octave development

conference this week, and some people made the suggestion of using

HDF http://en.wikipedia.org/wiki/Hierarchical_Data_Format as a modern

data container format.

I'm not suggesting we change, but we should at least think about HDF

and have some clear reasons about why we don't want to use that format.

--

Andy Adler <ad...@sce.carleton.ca> +1-613-520-2600x8785

[Pascal Olivier Gaggero]

Dear Andy,

thank you very much for your comment and idea. I recognize that HDF format may have some clear advantage and if I am not mistaken Matlab do also make use of it to store its .mat files. The argument not to use HDF, at least for this first version (i.e. version 1.x) of OEIT, would be the one of the broadest compatibility. What I find very attractive in using ZIP encoding is that you now-a-days find ZIP codec installed per default in almost every OS. The choice of using ZIP was also made by .docx and .odt developers. I guess that the main reason for that is that you can browse the archive as folder/file structure, which I find very convenient.

The point is that ZIP is not a definitive choice, if at some point HDF becomes a much broader standard we could switch to it any time because the basic structure of OEIT remains the same and is not based on the way we encode the data. This was the idea behind decoupling the content (folder, files, configuration files...) from the way it is actually stored in binary format on the hardrive.

What do you think? What would be the main argument to switch to HDF?

Pascal

[Andy Adler]

Hi Pascal,

Thanks. I think this is a good answer. We do need to be able to give a
reason to those who ask why we didn't choose the HDF format.

--
Andy Adler <ad...@sce.carleton.ca> +1-613-520-2600x8785