[arc-discuss] sacMaiL Error [PSARC]: Re: Discussion requested for a new ARC Policy on "Expectation Taxonomy and Repository Requirements"

[John Plocher]

[Please - arc-discuss ONLY.  Do not Cc: PSARC or PSARC-ext, this is NOT
case review material.]

Garrett D'Amore wrote:
> * namespace collisions... I'm not sure what level, if any, IPS has to 
> separate out namespaces for different "expectation levels".  I presume 
> we are talking about separation of installable packages -- conflicts in 
> actual *installed* bits are allowed, right?

Yup. Nothing in this proposal deals with installed name or filesystem
location details.

I thought about (and rejected) something like ...

   If I ask for the "Core" level, I'd better NOT get anything from
   one of the other levels, even if nothing in Core has the name I asked
   for and it exists at another level.  It may be OK for things to
   work the other way, though - asking for Foo in Experimental could
   well be satisfied by a package found with a Preferred tag.

... simply because it felt more like an implementation detail and not
an architectural one.


> * I do like how you have separated out the levels, although I'm not sure 
> there is a lot of merit in separating out Sandbox from Prototype.  (In 
> particular, more detail explaining what the difference between these 
> levels would be needed.)

I added:

# The purpose of the Sandbox tag is simply testing of the
   package building and repository interaction abilities.
   Sandbox expectations include a strong sense of transience
   - expect the repository to be periodically cleansed of
   "old" Sandbox tagged items.
# Experimental items are similar to Sandbox, but without
   the transient expectation.
# Experimental tagged projects are expected to be ARC
   "SelfReview" closed approved automatic fasttracks. They
   exist simply to record the package name and version.
# Prototype items carry an expectation that they will be
   superseded in the future by a package with a higher
   "quality" tag - either Preferred or Core - but at this
   point they are still Experimental.


> * In Project Behavior, I think it should be permissible for ARC and a 
> project to explicitly agree to demote (or maybe even promote) 

I added:

# During the ARC review process, the ARC and/or Project team may change
   the "expectation level" setting.


   -John