--- Stefan Bodewig <[EMAIL PROTECTED]> wrote:
> Hi all,
>
> I'm not really sure what the task is going to be
> used for, that's
> probably my biggest problem with it. Why do we need
> it?
For good or ill, when I was researching the
enhancement request, I noticed that AFS implements
Cloneable and it struck me that while additional
attributes and elements cannot be set on a reference,
it would be doable to apply additional configuration
options to a clone of a given AFS. In general, I tend
to solve problems as generically as possible, so my
natural inclination was toward enabling such use of
any Cloneable object. Nearly a month passed between
my declaration of intent to implement this idea and my
committing the task to CVS; unfortunately it must have
slipped through the cracks as bugrep comments often
seem to do and we now have a conflict. Had I known
the task would encounter such resistance I would have
saved myself the trouble of writing it. But it
happens sometimes...
>
> The problem Bug 32631 talks about could better be
> solved by tasks that
> perform set operations. Right now our best file
[SNIP]
Of the various ideas here, the path intersection stuff
sounds promising although I'm not sure it wholly
solves the issue of Bug 32631. I still think FileSets
with null dir attributes are valid, do not violate the
Java APIs, and stand little chance of breaking things,
and due to some arguments on set logic I am willing to
see nested filesets or some representation of that
concept (depending on how convoluted the AFS code
might become as a result of nesting filesets). Nested
filesets could provide an alternate syntax for what
clone solved with regard to filesets, basically
allowing more selectors/patternsets to be applied only
to the files returned from the nested filesets. There
was a long discussion on these topics recently on Bug
5035.
This looks like two threads.
-Matt
__________________________________
Do you Yahoo!?
The all-new My Yahoo! - Get yours free!
http://my.yahoo.com
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
