BIND 10 #50: Review data source and query logic

[BIND 10 Development]

#50: Review data source and query logic
--------------------------+-------------------------------------------------
 Reporter:  each          |        Owner:  jinmei                                           
     Type:  enhancement   |       Status:  assigned                                         
 Priority:  critical      |    Milestone:  02. Running, functional authoritative-only server
Component:  Unclassified  |   Resolution:                                                   
 Keywords:                |    Sensitive:  0                                                
--------------------------+-------------------------------------------------

Comment(by each):

 Replying to [comment:18 jinmei]:
 > I suggest we forget things under lib/dns in the context of this ticket
 and get them reviewed as part of the DNS library module.

 I had the impression these things had already been split out into other
 tickets (and in many cases reviewed).  I'm not sure this ticket needs to
 be open any longer.

 >  1. data_source.{h,cc} and query.{h,cc} (and their tests)
 >  2. sqlite3_datasrc.{h,cc} (and their tests)
 >  3. static_datasrc.{h,cc} (and their tests)
 >
 > I had substantially modified 2 and 3 while I gave them my review, so
 someone else should review these.

 I paid close attention to your modifications while you were making them.
 I'm not averse to going over it again, or having someone else do so, but I
 actually think it would be reasonable to consider these reviewed.

 >  - As I said somewhere (in another ticket or bind10-dev), I suspect
 we'll need to substantially revisit the whole design and implementation of
 this stuff (as well as the main auth server code in src/bin/auth) anyway.
 Assuming we agree on this, perhaps we can simply complete a lightweight
 review just for the next week's release, and restart the whole work mostly
 from the scratch.

 I don't think I agree with "from scratch".  After the initial release I
 went over the current code with an eye toward redesign, and I actually
 think what we've got isn't bad.  Its major weaknesses are performance
 (attributable to slow and repetitive SQL queries and no cache), glitches
 with zone cuts, and of course non-writability.  Some new API will be
 needed for writing to the database, and a redesign of the existing API
 might help with the zone cut issues, but the general approach we're taking
 still seems fairly reasonable to me.

-- 
Ticket URL: <https://bind10.isc.org/ticket/50#comment:19>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development