> The JDK IP address parser will accept [], I believe, but it's
not a
> valid address.
Yes; the reason -b [::1] works is because InetAddress.getByName("[::1]")
works. It's not a case of the AS exerting effort to allow it to work;
we'd have to add code to make it not work.
gotcha.
> I'd say that we should allow [] but not recommend it.
Agreed. If the AS stops supporting it we will break people who for
whatever reason have needlessly added the []. If we do that we need to
have a good explanation (almost a FAQ) as to why we made that choice.
yup.
> For tools, it's a question of parsing the IP address
(getAddressByName
> should do it), determining if it's a nameless IPv6 address, and if so,
> escaping it; not too onerous a burden in my opinion.
not sure where that getAddressByName is ? you mean getByName(str) from
http://docs.oracle.com/javase/6/docs/api/java/net/InetAddress.html ?
But that is going to do a dns lookup - and those are damn slow if they don't exist on
some OS's.
Now that I've realized ipv6 is pure evil from a usability side the the way I'm
thinking is that we will
assume the address is "raw" and if we need to create a url we will do this:
public static String formatPossibleIpv6Address(String address) {
if (address == null) {
return address;
}
if (!address.contains(":")) {
return address;
}
if (address.startsWith("[") && address.endsWith("]"))
{
return address;
}
return "[" + address + "]";
}
and if we need the raw address we wont do the above and then it depends on the lower
layers
to treat it nicely ;)
lower layers here means parts of eclipse we can't change to do these kind of escaping
or unescaping which
might be necessary in some places too.
/max
>
> On 05/25/2012 06:47 AM, Max Rydahl Andersen wrote:
>> so in your future using [:::1] should error if not used in a url ?
>>
>> That is what ssh command line seem to be doing btw
>>
>> /max
>>
>> On May 25, 2012, at 13:43 , Pavel Janousek wrote:
>>
>>> Hi Max,
>>>
>>> I really hate to use [xyz] anywhere I'd like to use literal IPv6 address.
IP address should be always without brackets! On the other hand, URL need brackets because
URL port (+ password) delimiter is the same as IPv6 IP's group delimiter - in this
case [] are needed in _every_ case, but again - only when we are using _literal_ IPv6
address, not name. In an URL brackets are important to use because ambiguity can be
occurred very easily.
>>>
>>> I hate the present (only one) possible option as ./standalone.sh ... -b
[IPv6] -bmanagement=IPv6 because it is really bad and at least unprofessional... in every
case we'd like to specify literal IP address, we should do this _only_ without
bracket, because this is only one possible text representation...
>>>
>>> I guess, my thought isn't accepted by majority up to now (as we can see
in present AS7), but I'd like to bet future is mine...:)
>>>
>>> Pavel
>>>
>>> ----- Original Message -----
>>>> From: "Max Rydahl Andersen"<max.andersen(a)redhat.com>
>>>> To: "jboss-as7-dev(a)lists.jboss.org
Development"<jboss-as7-dev(a)lists.jboss.org>
>>>> Sent: Friday, May 25, 2012 12:30:44 PM
>>>> Subject: [jboss-as7-dev] ipv6 "usability" question
>>>>
>>>> Hi guys,
>>>>
>>>> Trying to be sure our tools actually can connect/work with ipv6
>>>> enabled servers we're are hitting a bit of a snag.
>>>>
>>>> Is it expected that clients unescapes/escapes ipv6 addresses or is
>>>> :::1 and [:::1] considered equivalent ?
>>>>
>>>> i.e. do code need to add []'s when constructing urls, and remove
[]'s
>>>> when passing it to things that just need an address ?
>>>>
>>>> Just wondering if we could tells users to put [::1] into hostname
>>>> fields if they want to use ipv6 - then we don't need to add all this
>>>> escaping/unescaping
>>>> for ipv6 in the UI layer of tools.
>>>>
>>>> /max
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> jboss-as7-dev mailing list
>>>> jboss-as7-dev(a)lists.jboss.org
>>>>
https://lists.jboss.org/mailman/listinfo/jboss-as7-dev
>>>>
>>
>>
>> _______________________________________________
>> jboss-as7-dev mailing list
>> jboss-as7-dev(a)lists.jboss.org
>>
https://lists.jboss.org/mailman/listinfo/jboss-as7-dev
>
>
--
Brian Stansberry
Principal Software Engineer
JBoss by Red Hat
_______________________________________________
jboss-as7-dev mailing list
jboss-as7-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/jboss-as7-dev