Both links will be "evaluated" (if by "evaluated" you mean processed by an
XLink-aware processor in order to enable traversal). When one gets used
depends on how the user interface enables traversal. For example, if you
click on the presented representation of the inner link you might go
directly to element "c" or you might get a dialog giving you a choice of
which link to traverse. But not that this is *entirely* dependent on how
the user interface implements traversal initiation, which the XLink
specification says nothing about (nor should it).
If you traverse the outer link and the replace show action is respected,
then you would lose the ability to traverse from the inner link unless you
did a "back" following the first traversal.
This all assumes that actuate="user". If actuate="auto", then all bets are
off, because there doesn't appear to be any obvious way to satisfy both
actuate requests in any meaningful way. If I was implementing it, I'd
probably say that the first link in tree order governs, which means the
outer link would be actuated and the inner link would never get a chance to
actuate, because the active document would be DOC1 as soon as the actuation
occurred.
This suggests that actuate="auto" is a dangerous tool that should be used
carefully.
Cheers,
E.
-- <Address HyTime=bibloc> W. Eliot Kimber, Senior Consulting SGML Engineer ISOGEN International Corp. 2200 N. Lamar St., Suite 230, Dallas, TX 75202. 214.953.0004 www.isogen.com </Address>