[bind10-dev] The #356 and #408 branches

[Michal 'vorner' Vaner]

Hello

On Tue, Dec 07, 2010 at 06:32:50PM +0100, Michal 'vorner' Vaner wrote:
> On Tue, Dec 07, 2010 at 10:44:38AM +0800, wanghaidong wrote:
> > It's OK for me.
> > 
> > Today, I will make some minor changes as Stephen's review feedback, then we can start to merge.
> 
> I somehow planned to have it reviewed a little bit first. But I don't mind
> fixing the review comments after the merge either, so it is OK with me this way
> as well.
> 
> I added most of the tests and finished most of the TODOs I had today anyway. OK,
> then, I'll read your code and try to merge it (it seems your side has less
> changes, so it will be less work for me to read yours and merge it, while I
> remember mine that the other way around).

I started to sync mine branch with yours (so I do not break your work in the
process). I commited what I have right now, but it is work in progress. It does
compile, but the tests do not run and a lot of code needs to be either
uncommented or updated. If you want to have a look, you can, I hope to finish
the sync on Friday (I'll not have time tomorrow I guess). The current work is on
top of branches/trac408 (but I accidentally sent it to the server as two
commits instead of one).

There are some notes, what needs to get updated:
• Some of my code still uses the one list of addresses, so we are not compatible
  there yet. That one should not be a problem.
• I defined an enum type for the address family, the current short didn't look
  right (type safety, the hacks that neither of AF_INET, AF_INET6 is null, need
  for the OS to define them, and I wanted to index by the value, which I can't
  do, because the values can be far away from each other). That update should be
  simple as well.
• While in the svn merge, I added locking to many places. From the deadlock in
  test, it seems I added them to too many, maybe I didn't read all the code, so
  there might be some missing.
• We probably have small design conflict there as well. You added the address
  selection into NameserverEntry. However, zone might have (and should have)
  multiple nameservers, meaning multiple NameserverEntries. The selection should
  be done over the union of all their addresses, not pick somehow one nameserver
  and then choose only from its addresses. I have a place marked in my code
  where I think it should happen (where I simply use some kind of random to just
  pick something). I hope it will be easy to move it somehow the one level up.

Have a nice day

-- 
This email has not been checked by an antivirus system.
No virus found.

Michal 'vorner' Vaner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: not available
URL: <https://lists.isc.org/pipermail/bind10-dev/attachments/20101208/4dc50ccb/attachment.bin>