Odd problems trying to make use of libbind as a replacement resolver...

[Brad Knowles]

At 6:31 PM +1000 2005-10-18, A Humble Bind User wrote:

>                             One casualty of "Open Source" is the 
>lack of formal
>  design documentation in advance of the first line of code.  The Sleepycat DB
>  library as Paul mentioned is a model exception of course, but most 
>of the time
>  "read the source" is the rule of the game when it comes time for
>  "documentation", which is just plain awful.

	With regards to BIND itself, there is a very large amount of very 
high quality documentation available, most of it from Paul Albitz and 
Cricket Liu, in their various books on the subject.

	However, when you're talking about the resolver library, that is 
systems programming territory, and pretty much the exclusive purview 
of the individuals at the vendor who is maintaining that code.  And 
they had damn well better be reading the code itself and know 
backwards and forwards what it is really doing inside, as compared to 
whatever the documentation claims.

>  I'm not asking for anyone to "solve my problems for me"...  After 
>15+ years of
>  development, and pretenses of a "standardised UNIX platform", I would have
>  thought such a core component would be converging towards some standardised
>  behaviour.

	When have you ever known anything in this industry to move 
towards a "One True Model" of anything, especially in the open 
source/free software community, and most especially when that 
community is responsible for maintaining current code where previous 
generations were cannibalized and incorporated into systems libraries 
in virtually every OS on the planet?

	Can you even point to a single instance where this has happened? 
You can't use gcc as the model, because pretty much all the vendors 
who use gcc contribute their work back to the FSF, and use the 
version provided by the FSF as their standard system compiler.  You 
can't use sendmail as the model, because there are now a wide variety 
of MTAs available, and again most OSes that use sendmail take the 
standard version as provided by the open-source project.


	BIND is the only case in the world (that I know of) where this 
has happened.  And each vendor has clearly gone off on their own 
separate route afterwards.  It's simply not possible to reverse that 
process.

>  All said, it was a hope borne of some encouraging signs that the BIND9
>  project was a stab at "cleaning house" and building a name server
>  infrastructure that is maintainable and extensible without risking
>  disaster after each new feature.

	On the server side, this has happened.  On the library side, 
you're talking about code that is too deeply embedded in virtually 
every OS on the planet, and you can't just casually rip that out. 
Heck, you can't even come up with a general-purpose replacement, 
because each vendor has extended that code in their own (usually 
proprietary) way.

>  I was asking questions, not demanding solutions.

	That's not the impression I was getting.

-- 
Brad Knowles, <brad at stop.mail-abuse.org>

"Those who would give up essential Liberty, to purchase a little
temporary Safety, deserve neither Liberty nor Safety."

     -- Benjamin Franklin (1706-1790), reply of the Pennsylvania
     Assembly to the Governor, November 11, 1755

   SAGE member since 1995.  See <http://www.sage.org/> for more info.