On Feb 11, 2008 6:55 AM, Ivica Brodaric <[EMAIL PROTECTED]> wrote:
> And maybe extract the comment via QUERY MDISK?
The mistake of having QUERY MDISK USER as a class G command would then
make this even more attractive to those who have no business need to
know... ;-)
Sorry to spoil the party, but I think implementation of the entire
thing is less trivial than it appears. Such new statements also
require delegation in DIRMAINT. Some people may be authorized to
adjust the comment to their directory statements, but not the actual
statement. And should it change when the statement changes? Old
documentation often is worse than no documentation ("But it said *old*
version" / "Yes, that was before I put the new stuff there")
A popular attempt to documentation has been to move the actual MDISK
statement to another placeholder userid. This is limited, but can be
enough to classify stuff. The more specific the comments, the more
likely they will be out-of-date.
Another approach we've used in the past is a common file on each
system disk that lists the purpose and change history (e.g. @WHATSON
THISDISK). As anything else it requires discipline to keep it valid.
Or if you want, one single disk in the system that has one file per
system disk, named with userid and CUU.
If you have the information in a file or multiple files, it is very
easy to use a lookup stage to annotate some output with this (I do
this also for SFS directories, for example - how about adding comments
there as well).
Obviously with competition in this area, it takes a very long time
before things change in the directory. IMHO that is a good thing. I
have at least a dozen other things I would rather see VM development
do.
-Rob
