(Note: I don't favour one proposal over the other - I haven't spent enough time yet to think about the issue in depth. In general I'd vote for the "leanest" solution.)
one quick comment though:
> I can't see the benefit of having a list of images.
Images might be hidden, locked or otherwise not accessible (referenced in the <defs> section? what happens with clones?) , especially if you deal with documents you didn't create yourself (the case where image links are most likely to break) and therefor don't know it's internal structure.
(Note: I don't favour one proposal over the other - I haven't spent enough time yet to think about the issue in depth. In general I'd vote for the "leanest" solution.)
one quick comment though:
> I can't see the benefit of having a list of images.
Images might be hidden, locked or otherwise not accessible (referenced in the <defs> section? what happens with clones?) , especially if you deal with documents you didn't create yourself (the case where image links are most likely to break) and therefor don't know it's internal structure.