On 03/09/09 12:37, Dave Miner wrote: > Jack Schwartz wrote: >> Hi Dave. >> >> On 03/09/09 09:04, Dave Miner wrote: >>> Jack Schwartz wrote: >>>> Hi Dave. >>>> >>>> After coding up the changes per RFC 3696, it occurred that in the >>>> past I've used an underscore in the name without problems. I'm not >>>> sure that's the right document. >>>> >>>> Meantime, I've found out that what this is used for is a multicast >>>> DNS address. There is a document: >>>> http://files.multicastdns.org/draft-cheshire-dnsext-multicastdns.txt >>>> >>>> which says what can and cannot be in a multicast DNS address in >>>> section 18. >>>> >>>> In the end it says: >>>> "Multicast DNS implementations MUST NOT use any other encodings >>>> apart from precomposed UTF-8 (US-ASCII being considered a >>>> compatible subset of UTF-8)." >>>> >>>> So does this mean that any char (value 255 or less) is OK? >>>> >>>> If so, maybe checking only for dots to reject is OK... >>>> >>>> If I'm confused, can you please shed some light. You seem to know >>>> a lot about this... >>>> >>> Not really any more than has been brought to light here, I just >>> happened to be aware of IDN support coming in some time ago. >>> >>> For the moment, assuring that it's in the ASCII set is probably >>> sufficient. Full support for IDN appears to require work in the >>> applications to convert from non-ASCII to ASCII forms [1]. >> OK, so it sounds like I'm back to restricting to no dots but all >> other ASCII characters (numeric value <= 127) are OK. >> > > Printable, yes. <= 127 includes non-printables. If this is in C, isprint (is your friend!)
-Sanjay > > Dave > >> Thanks, >> Jack >>> Dave >>> >>> [1]http://en.wikipedia.org/wiki/Internationalized_domain_name >>> >>>> Thanks, >>>> Jack >>>> >>>> >>>> >>>> On 03/07/09 13:49, Dave Miner wrote: >>>>> Clay Baenziger wrote: >>>>>> Hi Jack, >>>>>> I think this looks good, it should certainly make life more >>>>>> simple for our users. >>>>>> I do think checking for just dots in the service name is a >>>>>> little too narrow. I was looking at the section "Restrictions on >>>>>> domain (DNS) names" in RFC 3696 and believe we may instead want >>>>>> to check that there's only [a-zA-Z0-9_] in the name to ensure >>>>>> there's not unanticipated problems as currently it looks like I >>>>>> could get an ampersand or pound sign in which could be difficult >>>>>> for some of the installadm shell scripts and the DNS system too. >>>>>> Let me know your thoughts on this or if being C checking for dots >>>>>> is about all that can easily be done at this time. >>>>> Internationalized domain names are now supported in the internet >>>>> infrastructure, I don't recall the specific RFC that covers the >>>>> legal values, but likely needs to cover more than the range above. >>>>> >>>>> Dave >>>>> >>>>>> Thank you, >>>>>> Clay >>>>>> >>>>>> On Fri, 6 Mar 2009, Jack Schwartz wrote: >>>>>> >>>>>>> Hi everyone. >>>>>>> >>>>>>> Here is a code review for a couple of small bugfixes: >>>>>>> >>>>>>> 5091 AI install does not work if your service name had . in it. >>>>>>> 4610 most installadm commands need to err out gracefully if not >>>>>>> root >>>>>>> >>>>>>> http://cr.opensolaris.org/~schwartz/090306.1/webrev/ >>>>>>> >>>>>>> Please review. >>>>>>> >>>>>>> Thanks, >>>>>>> Jack >>>>>>> _______________________________________________ >>>>>>> caiman-discuss mailing list >>>>>>> caiman-discuss at opensolaris.org >>>>>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss >>>>>>> >>>>>> _______________________________________________ >>>>>> caiman-discuss mailing list >>>>>> caiman-discuss at opensolaris.org >>>>>> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss >> > > _______________________________________________ > caiman-discuss mailing list > caiman-discuss at opensolaris.org > http://mail.opensolaris.org/mailman/listinfo/caiman-discuss
