Header: Choice of file format

[Lestropie]

I think this was probably discussed to some extent in earlier threads, but it's worth separating into its own thread and reaching a clear consensus.

In the current specification proposal, YAML is used for the header file. I think I had commented elsewhere that since BIDS uses JSON, many neuroimaging softwares will already have in place the requisite libraries for reading & writing JSON, and therefore using YAML instead may unnecessarily introduce a new dependency. From memory the primary argument for YAML was that it is supposedly more human-readable than JSON: I don't have a huge amount of experience with the YAML; personally I think JSON is "human readable enough".

Open to any and all opinions, or introduction of confounds that I've not thought of.

[jdtournier]

I was originally leaning towards YAML since it's supposed to be more human-readable, which I really value. Having read a bit more about the format, I have to say that it's actually a bit of a dog's breakfast... This article goes through a lot of the issues. One of the key issues: it's actually far too complex - reflected in this quote from the article above:

The YAML spec is 23,449 words; for comparison, TOML is 3,339 words, JSON is 1,969 words, and XML is 20,603 words.

So on reflection, I would also support using JSON rather than YAML, on the grounds that it's actually far simpler to parse and validate.

[frheault]

I will switch back to json. These are good point, I personally like json too. I switched at the time because you proposed it and it took 5 minutes. I will make the changes soon.

[MarcCote]

I recently stumbled upon Hjson (Human JSON). It supports comments and is more human-proof (e.g., deals with missing commas). I'm not advocating for it since I haven't used it yet.