multi-indexed tdb

[tridge at samba.org]

Jerry,

 > Simo,  Is this really necessary?

I can understand Simo flaming, as Volkers statements were extremely
inflammatory. It would have been better if Simo had ignored the
inflammatory remarks, but I can understand the immediate reaction to
flame. I had much the same feeling myself, and almost wrote a flame
mail, then thought better of it.

Volker made the following statements:

1) he said that Samba4 developers don't care about portability, and
   that we consider it as an afterthought. Given that it is Andrew and
   myself that have done 90% of the work on the build farm over the
   last 5 years I found this accusation grossly unfair. 

2) he made the general claim that our code is unstable. Compared to
   what? On what basis? It's the sort of silly statement. You can
   always make this type of claim about any code base. If its being
   actively developed you call it unstable. If its not being actively
   developed you call it abandoned. Just like someone who wants to
   paint a bright picture does the reverse, and says that code that
   hasn't changed is stable and code that has been changing is being
   actively maintained.

   I also think this statement is grossly unfair given the huge effort
   we have put into test suites and testing over the past few years,
   and given the benefit that all branches of Samba have received from
   that effort.

3) he claimed that ldb does not compile standalone, is intertwined
   into Samba4, and used this to dismiss it. That is not true at all,
   and seems to have arisen because Volker did not know what popt
   is. We've gone to quite a lot of effort to ensure that ldb is
   usable outside of Samba4.

I'd be delighted to get some technical debate on testing, code
stability, interfaces etc, but the inflammatory generalities that
Volker made are not what I had in mind.

 > For the record, talloc_steal() was banned in SAMBA_3_0
 > based on Jeremy's concerns.

I'll be intererested to discuss that with Jeremy. Maybe I missed the
discussion on that decision? I have not been keeping up with email at
all well lately.

Maybe my 'about talloc_steal()' email will help kindle some debate on
talloc_steal() and whether it is a good interface?

Cheers, Tridge