On Tue, 30 Aug 2005 23:14:40 -0400 Michael Jennings <[EMAIL PROTECTED]> babbled:
> On Wednesday, 31 August 2005, at 11:59:43 (+0900), > Carsten Haitzler wrote: > > > how many people do dns via nis/yp or ldap? sure - nice theory - but > > actually do dns VIA that? i knwo dns servers are oftne BACKED by > > ldap as such - but other than very specific special cases in niche > > areas - actual dns VIA nis/yp or ldap... i don't know. > > It's not DNS via NIS/YP. It's name resolution via NIS/YP. It's well dns == dynamic name resolution... i was being loose with my terms i admit :) > really no different than using /etc/hosts. Just like /etc/passwd is > the local file which can supplement NIS or LDAP account info, > /etc/hosts can supplement NIS/LDAP host resolution. And yes, it's > done, usually in Windows-centric (Active Directory) setups that want > to use WINS and ADS without having to set up DNS. ugh! ok - point taken - i dont live in any of those setups... ever! :) > > AFTER seb had written the dns procotol handling that came to mind and i > > suggested it... but i have to say i respect seb's work on doing th edns > > protocol handling and reading the rfc's etc. and i dont see just nuking the > > code as good. i am leaving it in seb's hands for now. it needs completeness > > handling. i dont see ldap/nis/yp as usefully worth implementing and if you > > happen to have a system with these setups - then fall back to blocking > > gethostbyname is an option :) something about a librarye forking off child > > processes seems a bit evil to me - and thus i hesitate on it. but then again > > re-implementing dns is evil too :) > > Yeah, it's a lesser of two evils thing. The problem with the current > method is that the standard dictates checking nsswitch *before* doing > the lookup and potentially not using DNS at all. Honestly I think > forking a new process/thread (just fork(), no exec(), maybe vfork() if > possible?) is cleaner than trying to re-implement not only DNS but > nsswitch too. thats true - the fork will be really lean as its copy on write only a small stack segment will be copied over. and yes - i admit - we need to honor nsswitch etc. etc. etc. if doing dns DIY. :) > > that's possible - but its a service that has to sit around and be > > managed. i'd much sooenr go with the fork off a child to dns lookup > > via gethostbyname then punt the retuned data via fd back to the > > parent method long before this :) > > Fair enough. :) it's all a matter fo levels of evil i guess. god damn why did they not just impement a proper async dns lookup to start with in libc! ARGH! fools! :) -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- The Rasterman (Carsten Haitzler) [EMAIL PROTECTED] 裸好多 [EMAIL PROTECTED] Tokyo, Japan (東京 日本) ------------------------------------------------------- SF.Net email is Sponsored by the Better Software Conference & EXPO September 19-22, 2005 * San Francisco, CA * Development Lifecycle Practices Agile & Plan-Driven Development * Managing Projects & Teams * Testing & QA Security * Process Improvement & Measurement * http://www.sqe.com/bsce5sf _______________________________________________ enlightenment-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/enlightenment-users
