I tried to build a collection list of dataset pairs (paired end reads) and wonder if I just do it wrong, or if there is some intuitive functionality missing - tested on latest dev version:
I thought if I filter the left and right tables to the same amount of files, it shouldn't require much logic to just combine them with Auto-pair - however, this doesn't work:

To avoid other conflicts, I filtered even more, with the same result:

Is this intended, am I using it wrong? I ended up clicking Pair these datasets around 500 times, which worked better than it sounds, seems like it queues the clicks (finally putting my esport skills to good use).
I tried your example, it seems it just doesn't reach the score threshold for matching ... I was able to set https://github.com/galaxyproject/galaxy/blob/dev/client/galaxy/scripts/mvc/collection/list-of-pairs-collection-creator.js#L208 to 0.6, recompile the client javascript (make client) and then your example worked. Alternatively you can also increase the length of the input dataset names ... which doesn't make a lot of sense as a workaround.

@jmchilton do you think setting the score threshold lower is a possibility ? Or perhaps we could have a slider in the collection builder and expose this to the user with a warning that there could be mismatches ?
Yes, with 0.6 it works for me too - sadly I cannot change this value on the productive instance. I could see three solutions:
Pair these datasets for all - so that the user has to make sure he used the right filtering but then doesn't have to click it multiple times@mvdbeek I'm fine with lowering this value - sounds like a good idea.
Since this is quite a useful change, could we add this to the 17.09 milestone?
I'll open a PR for this right now.
Thanks!