Any need or interest for self-descriptive format?

[m_dr_m]

Hallo.


I'm polishing Ayane's "module" on-disk format, and was wondering whether it would be of any interest to use a self-descriptive format.
Not XML-like, god forbid. Rather IFF like, with tagged chunks + some common meta info for each chunk (version, size).


Recently Gunhed was interested in decoding Orgams files.
But in both cases a generic format appears neither sufficient nor necessary!


What do you think?

[norecess]

I think IFF is amazing because you can extend it as you wish, while remaining small data-wise. I used it for reading LightWave's LWO objects a long time ago. 30 years after, they almost reinvented IFF as "BSON" ("Binary JSON", http://bsonspec.org/).

On the other side, you can't directly read a fixed structure with IFF; you need to parse each block separately and recursively. This may slow down loading + much complicated to parse in asm.

Is your data fragmented a lot (with many small data structures) ? then yes, maybe it worth it.
If your data is like 1 or 2 "big structs" + some 2 or 3 linear data with variable size, well, maybe it does not worth it.

[GUNHED]

What do you think?

Do it the way it's best for Orgams, that keeps it quick.    However to know how the format works would be of advantage for other apps in case they like to show an .o file for example. To be realistic, on CPC there not so much coders. So it's worth thinking if the effort makes sense or not. 

[m_dr_m]

What I have for now: http://orgams.wikidot.com/ayaneformatspec