Have had some debates with both Aslak and John on the subject. From IRC today:

"#jboss-dev" wrote:
baileyje> ALR: ping
ALR> baileyje: Pong
baileyje> ALR: So I had thrown together a MemoryMap impl earlier before I realized Aslak had already put a partial one up.. Whoops
ALR> baileyje: Hehe, whoever gets there first in working form wins.
baileyje> ALR: he ran into the same issue I did. Asset does not have enough data to really support the Archive.add(Asset...) functionality.
ALR> baileyje: Yep.
baileyje> ALR: Is that encompassed in another task, or should there be an new task to update the API to support getting Path like info from an asset?
ALR> baileyje: This is something we'd debated a lot.
ALR> baileyje: I think I left it to aslak to determine if we need some getPath from Asset or not. Hoping we wouldn't need it.
baileyje> ALR: Yeah. Seems like you could get rid of it if required a Path of every addition
ALR> baileyje: It may be demanded by TMPARCH-6
ALR> baileyje: Right.
ALR> I think my proposal was that every Asset addition needed a path
ALR> And that "container " additions could know how to supply it.
baileyje> ALR: That would keep Asset cleaner
ALR> Like: addManifest("String") would assemble the default Path and Asset to add, then delegate to add(Path,Asset)
ALR> baileyje: Yes, but I'm not sure if it works for every case. I like the cleaner approach too.
ALR> aslak disagreed on that poing.
ALR> *point
baileyje> ALR: Yeah..
baileyje> ALR: He wants the default path as a required attribute on each asset?
ALR> baileyje: Yes.
baileyje> ALR: hmmm..
ALR> baileyje: I'll review your patches in a bit. Trying to keep up with you guys while I'm pulled away to other things this week.
baileyje> ALR: seems like we should require the path and leave asset clean and rely on the containers to establish the paths, until we have compelling evidence the Asset needs a path attr..
ALR> baileyje: That's my position.
ALR> baileyje: If you have the time to prove that to aslak, awesome.
baileyje> ALR: K. I will see if I can put something together.. It makes the MemoryMap much easier to impl a path required..
ALR> And if in trying to prove it he was right all along, we eat our words.
ALR> I'll mail him this little discussion
baileyje> for sure..
ALR> Or better, post on the forum.

Aslak, now public for your review at your convenience.

Also I'd like to do more forum posts for topics that should leave some trail. So we can all be in the conversation and remember the history.

S,
ALR