Ralf, there's only one reason why I mentioned moving things to a follow-up task: the desire not to block a particular code area for days to come. As a matter of fact, I am not happy at all about the patch, as I had to introduce code to CollectionInfo (if I remember correctly) that I do not like at all.
In other words, whilst the patch removes all the conditionals from XSList (by introducing XSNMTokens and XSIdRefs), it makes things worse in another area. Rather than pondering fro days about a better solution (that in my opinion would require a lot of refactoring), I was thinking about releasing a partial improvement earlier (allowing you and Eddie to merge changes in as they become available, and as a result be able to touch other code areas), and subsequently discuss the 'botch' introduced to CollectionInfo. As you have said yourself, we are not talking about a blocker, but about general refactoring to improve the code base.
In addition, I am rather in favor of making a patch ready for review early (to, again, get your feedback, e.g. about the approach taken), especially when knowing that I am about to leave office and that I won't be on-line for the remainder of the evening.