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

[A Humble Bind User]

> Red Hat Fedora Core(FC) FC-3, FC-4 and Rawhide ( FC-5 )
> all ship libbind, in the optional 'bind-libbind-devel' sub-package,
> the BIND 8.4.6? resolver library that comes with BIND 9.3.1 / 9.2.5.

Ah, that's interesting...

> But BAD problems result, similar to / including the ones you describe,
> if you just 'make; make install' in the default source lib/bind directory,
> because this WIPES OUT your glibc's resolver header files in

Oh dear, yes... 

> So the motto is:
> NEVER MIX COMPILES AND LINKS BETWEEN libbind AND THE glibc resolver

I'm fine with that, really.  My particular problem was that something like
pthreads had some linkage with libresolv, which meant any multi-threaded
library would lock me into libresolv unless I rebuild all dependents w/o
threading support.  This isn't too onerous or painful really, as many
workhorse UNIX tools aren't threaded at all (sshd, apache-1.3, sendmail,
uw-imap, etc).  

I already segregate OpenSSL builds, and up until semi-recently, was content to
let the "divide" be shared versus static, as the really security/performance
conscious things I build statically (on Intel this returns a precious register
to common use).  Sure, it's an incremental improvement, but these increments
do add up (the threaded code exacts a small performance penalty as well).  But
this is turning out to be insufficient.

> The glibc resolver as shipped in Linux glibc-2.3 is baselined off the BIND
> 8.2.3 libbind,
> NOT BIND 4.x, with significant changes to improve multithreading and IPv6
> support .

That doesn't sound too bad, then.

> Yes, it does need upgrading to support DNSSEC and possibly
> internationalized domain names, and work is underway to do this.

Heh, well, that's overdue.  I wonder if it'd be possible convince Herr Drepper
to bring some harmony in libresolv with current "modern" sub resolver
libraries, even if it's impossible or impractical to keep it "isolated" so
that upgrading it isn't onerous.  As it is, at least pthreads seems to be
compiled with libresolv's constants/etc.

I appreciate your informative reply.

=R=