On 05/ 5/10 08:48 AM, Liane Praza wrote:
On 05/05/10 06:59, Garrett D'Amore wrote:
I tend to agree that from an architectural perspective, the site.xml
might better be handled as Obsolete -- it seems that the new mechanism
is superior.
The new mechanism is identical, so the site.xml location simply
augments the site directory with the same semantics, and costs us very
little to maintain.
I don't see a real benefit in marking it obsolete committed at this
point, but if the ARC really wants that modification, it certainly
causes no change in the implementation, and we can trivially make that
change in the materials.
Generally, I don't choose the Obsolete Committed direction for
interfaces unless if we truly can't get rid of something but there's a
new interface with truly superior and new features. (For things we
get benefit from removing and can actually remove, I go straight to
EOF as quickly as possible.) In this case, the semantics are the
same, so there's no real benefit to encouraging customers to convert
if they're happy specifying a monolithic profile rather than one in
fragments.
My take on these things is:
1) If there is a clear preference of one interface vs. another that
customers should use (and the interfaces offer identical functionality
with no reason to use the less preferred), then we ought to make it
Obsolete.
2) Eventually, cleaning up duplicate code is a good thing.
3) Duplicate interfaces (like locations to store config data) can cause
future problems for other software, that has to be aware of all possible
locations (and may run into the same problems that created the rationale
for this case in the first place.)
All these make me believe we should mark the site.xml interface
"Obsolete" to discourage its use, and remove it when it is easy to do so.
IOW, Obsolete Committed for site.xml wouldn't be my choice, but it's
also not critical enough to quibble, and I suspect Tony will feel the
same. :)
Obsolete Committed seems unfortunate -- is there a reason this needs to
be Committed? It seems like this could be decommitted somewhat for the
Solaris Next release? Is this file used on Solaris 10? (I don't recall
the history behind site.xml.) If not, then we could just nuke it now.
- Garrett
_______________________________________________
opensolaris-arc mailing list
[email protected]
