[bind10-dev] request for pushing a new task to the sprint (and middle term plans)

[JINMEI Tatuya / 神明達哉]

The following are my suggestion to Shane when we talked about task
plans for the beta.  I specifically propose pushing #2497 to the
current sprint (it's new, and not estimated).  Is it okay?  If it is,
do we want to give an estimation?  (I'm basically assuming it's okay
to push it to the sprint, so I'll do it unless I hear explicit 
objection).

---
JINMEI, Tatuya

I'd suggest a bit different approach for RDATAs: defer most of them
post beta except for those that have "character-string" (i.e., "text")
fields.  They are TXT, SPF, HINFO, and NAPTR.  For the rest, introduce
a wrapper factory that simply forwards the task to the existing "from
string" factory.  I believe this should cover the vast majority cases
(it may still miss some tricky broken cases, but should accept the
vast majority of valid input).  We need a new task for the wrapper,
though, and I suggest we push it to the current sprint (ideally after
estimating it, otherwise just give some intermediate value of "5").
I've created a ticket for it: #2497.  I also noticed TXT and SPF are
mostly covered in #2442, so I propose we complete the support in
#2442.  For HINFO and NAPTR we'll need a new ticket, which is #2498.

Another minor note: #2381 is not urgent and can be deferred.

So the minimum set of "loader" part will be:

Total points: 53 + #2497 + #2498 (60+-ish)
#2374 	MasterLexer's Number state classes	4
#2377 	define dns::MasterLoad class	7
#2378 	define and implement datasrc::ZoneLoader class	4
#2379 	add python wrapper of datasrc::ZoneLoader	4
#2380 	revise b10-loadzone using datasrc.ZoneLoader	8
#2382 	introduce generic version of rdata::createRdata(text)	5
#2427 	support $ORIGIN in dns::MasterLoader	3
#2428 	support $INCLUDE in dns::MasterLoader	3
#2429 	support $TTL and TTL guessing in dns::MasterLoader 3 
#2431 	support flexible ordering/appearance of TTL, type and class in dns::MasterLoader	5
#2442 	update TXTLikeImpl with from lexer constructor	4
#2470 	use MasterLoader in memory::ZoneDataLoader 	3 
#2497 introduce wrapper version of "from lexer" rdata factory
#2498 support generic version of rdata::createRdata(text) in HINFO and NAPTR

I think this is a realistic base line.  We plan to do 30 points in this
sprint, and I don't think we can hope much more than that in the next
sprint even if some people won't have holidays.

The next level is post-load validation.  I noticed one missing task
(#2499), and this is the entire list:

Total point: 16 + #2499 (20-ish)
#2432 	define and implement base and libdns++ version of RRsetCollection 5
#2433 	implement validateZone, base part	4
#2435 	implement datasrc version of RRsetCollection	4
#2436 	update datasrc::ZoneLoader so it performs post load checks using validateZone()	3
#2499 basic post-load validation for in-memory data source

I suspect we won't be able to complete these by the beta release, but
we can see at the time of the next planning.

Another possible next level is to support non-normalized input for
in-memory (directly from text).  They are:

#2440 	add support for "merging" RRsets into memory::RdataSet	6
#2441 	update in-memory data source so it can load RRs in any order	4
(total 10 points)