Cool 
We have that many? Okay! Checking, it seems we have 700-ish in English, some in Russian, German, and Indonesian, and over 3000 in Vietnamese.
Okay, good, now we know.
The general idea for legacy texts is that:
- Where a source text has reference IDs, these should be preserved.
- Where the source reference IDs follow a pre-existing system (eg PTS numbers) they should be labelled consistently.
We haven’t added IDs or attempted to systematize them. And we won’t: this is one of the things where it is better to just leave the past behind and focus on the new JSON system. But of course if things are broken or easily made compatible, that is fine.
data-uid is a special case that is used for texts that do not readily fit the “one text= one HTML file” pattern. This includes the ones and twos of AN and the Dhammapadas. The idea is that we can include multiple sets of IDs for handling different situation. To what extent they are actually used and properly applied I cannot say, perhaps @blake will know better.
It’s easy to add arbitrary anchors, we can just add an ID to <hX>, <p>, <ol>, <ul> elements that are currently lacking them. There will be no consistency between editions (which is already the case so that doesn’t change) but at least people can link to the bits they are interested in.