>There is no problem adding a new tag to PEERS NAMES as you suggest.
>However, the question is, what does this tag exactly mean? Do you want
>all the files that would be sent to a backbone server? Just PEERS NAMES
>and LINKSWT FILE? In that case, they would have to be moved to yet
>another filelist, as the level of distribution of a file is determined by
>the filelist it is listed in.
I definitely meant all the files that are sent to a backbone server. No
special subset - THAT would not make any sense.
>In other words, the question is the following: today, there is only one
>set of files which is distributed to backbone servers, and that is also
>the set of files that Doron wants. Tomorrow, they may be other files for
>backbone servers. How do we define whether these files are also part of
>the set that Doron wants or not? Do we need to distinguish between
>:backbone.YES files and :Doron.YES ( :-) ) ones?
1. See above about which files I suggested.
2. If this whole issue, as you imply, has only to do with *me*, then
I'd suggest to drop it immediately and stop discussing it - no
public interest. The suggestion for a new tag was exactly because
I thought this problem does not necessarily affect only myself,
and thus deserves a more general solution.
As for myself, I believe that :backbone.YES DISTRIBUTE(NO), as Eric
described it in a later note, will suffice. I just believe, that
subscription to the control files (again, the *complete* set) is
a facility that should be (a) easily obtainable and (b) detached
from all other facilities. (Of course, anyone who subscribes to
them is responsible for the outcome - namely, keep his LISTSERV
up to date).
Please remember that I've raised this whole subject in view of
DIST2 and not my personal wishes. Please note, that any DIST2
job initiated at node X on the network, if X is not on the backbone,
and its control files are backlevel, has a nasty effect on the
whole network and not just the particular Postmaster of the
particular server.
Regards
Doron
|
