[...] (see other parts of this thread for technical discussion)
>
> I could also setup a branch on the Subversion repository if anybody can hook
> my up with a commit bit there. My public key can be found here:
> http://dl.dropbox.com/u/1365846/tmp/id_dsa.pub
>
I think the information listed
hiya
I like the idea!
> thanks for your feedback. Good question. Currently it's defined by the
> behaviour of glob() which is bad, in my opinion. Maybe we can use the list
> of @classes and put a weight there. If you have this list for example:
> qw(CLASS1 CLASS2 CLASS3 CLASS4 CLASS5 CLASS6). Let
Hi Thomas,
On 25.11.2010, at 11:28, Thomas Lange wrote:
> I wonder what happens if there's a file CLASS4.CLASS5.CLASS6 and the
> client also belongs to those classes. Which file will fcopy choose
> then?
thanks for your feedback. Good question. Currently it's defined by the
behaviour of glob()
> On Thu, 25 Nov 2010 09:43:36 +0100, Johannes Plunien
> said:
> want, by dots. The most specific wins. So if there's a file
> CLASS1.CLASS2.CLASS3, it will win over CLASS1.CLASS2 if the
> server is in all three classes.
I wonder what happens if there's a file CLASS4.CLASS5.
Hi,
currently fcopy looks for files that are named exactly like one of the FAI
classes the server belongs to. Using my patch you can also add files named
CLASS1.CLASS2. So let's assume following situation: You got servers in CLASS1,
others in CLASS2 and even some which are in CLASS1 and CLASS2.