Interesting thread!
A bit of historical context to start with:
DROID version 6, in 2010 introduced Container Signatures, but many formats that went on to have Container Signatures already had binary signatures. Because TNA had no way of tracking user take-up of DROID 6, there was a decision that they should maintain backwards compatibility with earlier versions of DROID for a time, so this meant: a) not removing pre-existing Binary Signatures and b) I believe for a brief period they created both Binary and Container Signatures where this was possible. By the time I started working on PRONOM in 2012 the guidance was not to remove pre-existing Binary Signatures that also had Container Signatures, but IIRC I think I didn’t create any new ones where I was creating a new Container Signature.
My personal view on this now is that I think it’s been long enough that there won’t be anybody using DROID 5 or earlier in any formal context (maybe historical research/curiosity?) so perhaps it is time to remove the older Binary Signatures, but I also expect that they are mostly likely to be benign.
Here are all the file formats that have both at least one Container Signature plus at least one binary signature. Some big hitters in here:
fmt/39 - Microsoft Word Document 6.0/95I would hope that
in all cases the Binary and Container signatures at least agree on what they’re
seeking, or if they differ then the Container Signature contains greater
detail.
So in terms of clean-up – If these PowerPoint formats that have existing entries can get unambiguous identification, then great, please go ahead with that. If they cannot be disambiguated, but that means that version/date ranges & descriptions need to be tweaked then I think that’s good also.
If we do encounter any given instance where the binary/container signatures are fundamentally different then that instance will warrant revisiting – I expect in most cases the Container Signature will be more specific/accurate.
I hope this is useful,
David