Web Audio Evaluation Tool

As it is now (in the analysis/postprocessing code and elsewhere) the 'id' attribute of a question or an audioelement serves to retrieve them from result XML files.
The new XML checking requires that id's are unique for the whole document.

[Brecht] More often than not, though, I will have several pages which share mix IDs, e.g. when I have different mixes from different songs by the same six mix engineers - in this case, the fragment ID would be the same for mixes done by the same mix engineer. This allows me to calculate e.g. the average preference rating for a certain mix engineer, over different songs.
But even when the fragment ID is only the same for identical audio files (URLs), you could have several of those in the same file (on different pages). Consider for instance Giulio’s Hammond organ experiment or something similar, where one page (/test part) would compare a couple of ‘hard notes’ vs ‘soft notes’, and then a second page where ‘hard notes’ are compared to ‘medium notes’ - the hard note URLs, and logically also IDs, would then be the same.
Many other tests could be conceived where the experimenter would want to ‘reuse’ fragments throughout a test, particularly when only a random selection of pages is shown to each individual participant.

[Nick] The reason for requiring uniqueness is to ensure the id attribute meets W3 XML specification and behaves as an ID to uniquely identify a node. These ID tags are used to directly add pieces of information to the saves (it doesn’t right now but it will need to for intermediate saves, so we need unique tags). Plus I’ve already seen in the analysis output several nodes called McG-A a few times (or something like that) which would be quite confusing as at that stage.
Since all the files also have the URL/filename, it is very simple to extract if two fragments are of the same file, even in the results/analysis stages (given the URL by definition is unique to that resource).
So we have a method to extract the exact results node relevant to the exact page, plus a method to get the results nodes for the resource across all pages.

Resolution: let's add unique IDs but use 'names' for such questions or fragment identifier which would otherwise appear multiple times in a document. The analysis/postprocessing scripts should be modified in this regard. They could for instance check whether a 'name' is (also) used instead of an 'id', and give precedence to the name when plotting results, for instance.