> However, I am still puzzled by the purpose of it all.
> The new xr.sty is manifestly more difficult to follow
> than its predecessor.
is it really are are you only slightly taken aback by the many _ which
is used to typical latex code with @ signs is something one needs to
use two or three times to get used to.
if this is a problem for you try looking at it using a global replace
with _ to @
the example chosen by Richerd in the first place is perhaps not that
good as it is not allowing you to really make use of the features of
the language. the only think i used was the \seq module (sorry:-)
which actually also implements stacks so i could push and pop aux file
names onto the variable \l_xr_subfiles_seq and a few of these test
commands like \ior_eof:NTF which made the code a lot more compact.
in fact just by concerting it i found that i could do the package with
a much simpler algorithm at it is implemented right now.
> So who is meant to benefit ?
> The user, the package writer, ... ?
the package writer in the first place, the user indirectly as it
should lead to cleaner internal code which works better together.
> And what exactly is the benefit ?
as the example chosen does not really contain much code there is not
much benefit (other than those i tried to explain), for example all
the benefit about argument handling is not used.
but who knows, perhaps Richard (when the Europeans are sleeping) is
trying it once more on a more complex example (hopefully really trying
to use the additional functionality offered as well)
maybe there are others who wish to comment on the different versions?