Re: DNS servers
On Thu, Nov 21, 2002 at 03:34:43PM -0800, Nate Campi wrote:
> The problem with the BIND zone file format is that it's hard for
> programs to parse.
so, what you're saying is that it's too difficult for a program to strip
out comments and blank lines, or to join continuation lines (e.g. SOA
records) before parsing them. readahead is too advanced a concept for
programmers to wrap their heads around. i.e. to convert the input to
canonical form before (or as) it gets to the parser is too hard.
it's nowhere near as hard as you make out.
> So choice is great for people entering DNS info but bad for people
> writing zone file parsers.
IMO, a worthwhile tradeoff.
human readable & editable config files should be optimised for human
use, not for machine use. programs should do the extra work to convert
on-the-fly if required, not humans.
that's what computers are for, to make things easier for people, to do
the drudgework.
> Sure if you run BIND you can slave the zone to normalize the results
> and parse that - but OOPS there goes your comments and ordering.
yes, i've made that point. it's one of the reasons why zone-transfers
are not an adequate conversion tool.
> People who programatically manipulate DNS data often pick tinydns. Yes
> you can use BIND and dynamic updates, but many people find the high risk
> of update loss unacceptable.
parsing bind zonefiles is slightly more difficult than parsing tinydns.
*generating* bind zonfiles, however, isn't particularly difficult. no
more difficult than generating any other text file.
> This is about choice and the best tool for the job. Everybody shut the
> hell up and use what you think is best for you (and be glad you have
> choices in the first place).
i think you'll find i've made that point too. the trouble is that some
of the djb groupies find it intolerable that someone may have evaluated
djbdns and decided it wasn't for him. that's blasphemy!
craig
--
craig sanders <cas@taz.net.au>
Fabricati Diem, PVNC.
-- motto of the Ankh-Morpork City Watch
Reply to: