|
>> I would like anyone to help me to confirm if the following MCC
>> directories MUST be at the root level (.) or NOT,
>> DNA_NODESYMNONYM
>> MCC_*_BACKTRANSLATION
>> Thank you for an answer, and if possible a clear explaination for this
>> restriction.
The following directories must be at the DECdns namespace root level:
DNA_NODESYNONYM
DNA_BACKTRANSLATION
MCC_*_BACKTRANSLATION (where * translates to an MCC class name)
Reasoning:
DNA_NODESYNONYM these two are architected by the DNA
DNA_BACKTRANSLATION Phase V architecture
MCC_*_BACKTRANSLATION MCC constructs strings to lookup
instances in the DECdns namespace.
When you use an alternate identifier
these strings are used as the absolute
name. Because these strings are
constructed at runtime, MCC uses
'Well-formed' names, inserts
the class name into the backtranslation
directory name and appends the instance
identifier to find the soft link that
points at the object in the namespace.
...kjn
|
| To add to KJ's explanation...another reason for having a single
well-known backtranslation directory for each global entity class is to
ensure that there can only be one (and hence unique) name for every
global entity. Were we to allow multiple b.t. directories, we might end
up with more than one name for the same global entity.
Then...you get back into the old Phase IV problems, so when I say I am
dealing with (for example) bridge .foo, you may now it as bridge .bar
and we get muxed ip.
Colin
|