On Fri, 4 Sep 1998, Monte Ambrose wrote:

> We are looking into re-doing the way ADSM handles symbolic links.  I
> agree that it definitely needs changes.  Currently we are looking into
> making ADSM consistent with the way it backs up symbolic links for
> Archive and Backup.  One possible solution: Followsymbolic=yes will not
> follow the link for both.  Followsymbolic=no will not follow the link
> for both.  We do not like to add to many options but perhaps Archive and
> Backup should have their own option for Followsymbolic.  Does anyone
> have input into how this might affect them?  Currently if we follow a
> link and it results in a looping situation we detect it and we end the
> loop.  If you have a situation where you are getting into a loop because
> of this then you should pursue through technical support because it is a
> bug.

I'd be most happy with Followsymbolic=yes having the behavior of following
all symbolic links through to completion except in cases where a loop is
detected.   And then with Followsymbolic=no, it would be nice if it'd
archive the symbolic link as such and be able to retrieve it as a symbolic
link.  At least coming from a predominantly Unix background, those types
of behavior make the most sense to me.

As the client is currently implemented, there definitely appears to be
some buggy behavior.  The looped links I've happened across consistently
result in the archive process running until either the scratch media is
exhausted, or until the ever growing path names of the files associated
with the link exceed some limit in the path name length. I'll put together
some test cases and try them out with the latest clients for the Unix
systems we have.

-brian.
=========================================================================