Heh, you just beat me to the punch.
-Rob
On 07/10/13 17:34, Dmitry Samersoff wrote:
Rob,
Existing code uses if (JVM_GetHostName(myhostname, NI_MAXHOST)) so I
withdraw my last comments. Please, don't change it!!!
-Dmitry
On 2013-10-07 20:30, Rob McKenna wrote:
Thanks Dmitry! I'll correct
Rob,
Existing code uses if (JVM_GetHostName(myhostname, NI_MAXHOST)) so I
withdraw my last comments. Please, don't change it!!!
-Dmitry
On 2013-10-07 20:30, Rob McKenna wrote:
> Thanks Dmitry! I'll correct that nipick pre-push.
>
> -Rob
>
> On 07/10/13 16:47, Dmitry Samersoff wrote:
>> Rob
Thanks Dmitry! I'll correct that nipick pre-push.
-Rob
On 07/10/13 16:47, Dmitry Samersoff wrote:
Rob,
This version of your fix looks good for me.
Inet4AddressImpl.c:
Thumbs up.
Inet6AddressImpl.c:
Thumbs up.
173 if (JVM_GetHostName(myhostname, NI_MAXHOST)) {
Nitpicking, e
Rob,
This version of your fix looks good for me.
Inet4AddressImpl.c:
Thumbs up.
Inet6AddressImpl.c:
Thumbs up.
173 if (JVM_GetHostName(myhostname, NI_MAXHOST)) {
Nitpicking, explicit == -1 would be better here.
> Actually, can you tell me why you'd rather not
> include ipv6 loopba
Hi Dmitry,
Thanks a lot for the comprehensive review. W.r.t. line 210, I agree
there is a problem with the logic, but I'd like to suggest an
alternative solution:
- If addrs4 >= 1, then we will always be including loopback addresses in
the list. Since the idea of this extra condition is to e
Rob,
See below -
ll. 210 have to be fixed, other comments is just an opinion.
Inet4AddressImpl.c:
ll.132 - it might be better to move initialization to a separate
function, the same way as in Inet6 - I really like this idea.
Inet6AddressImpl.c
ll. 126.
it's better to move static int initia
Looks fine to me Rob, thanks.
-Chris.
On 09/20/2013 03:58 PM, Rob McKenna wrote:
After a brief discussion with Chris, we decided to revert the position
of the call to lookupIfLocalAddrs so as to give the local host primacy
over DNS.
Latest (and hopefully last) webrev here:
http://cr.openjdk.j
After a brief discussion with Chris, we decided to revert the position
of the call to lookupIfLocalAddrs so as to give the local host primacy
over DNS.
Latest (and hopefully last) webrev here:
http://cr.openjdk.java.net/~robm/7180557/webrev.03/
-Rob
On 14/09/13 00:00, Rob McKenna wrote:
Hi Bernd,
I should have said in the context of this bug. What I meant was removing
AI_CANONNAME doesn't resolve the issue as far as Mac is concerned. I.e.
I still see the UnknownHostException. In this particular case the
hostname is not set via the hosts file.
In the latest webrev the call t
Am 13.09.2013, 19:32 Uhr, schrieb Rob McKenna :
W.r.t. the use of AI_CANONNAME, this doesn't actually make a difference
in the context of this fix, but is definitely something that should be
looked at. I'll put it on the todo list.
I think it does make a difference: If you remove the CANON f
Hi folks, updated webrev at:
http://cr.openjdk.java.net/~robm/7180557/webrev.02/
Hopefully all of your concerns have been addressed.
W.r.t. the use of AI_CANONNAME, this doesn't actually make a difference
in the context of this fix, but is definitely something that should be
looked at. I'll p
Thanks for the comments folks. Chris, I like the idea of moving this
check below the GAI call. Dmitry, that loop is indeed a bit of a code
smell. I'll take care of it. Bernd / Dmitry, thanks for the notes on
AI_CANONNAME. I'll adjust the code and get some testing done then report
back!
I've k
Rob,
Instead of iterating interfaces list twice you can maintain two indices,
as you already know exact number of addresses.
put_v4_at(i)
put_v6_at(j+v4count)
Also you can take a look at
solaris/native/java/net/NetworkInterface.c
and possibly re-use some logic from there.
-Dmitry
On 2013-09-
Hello,
I reported before, AI_CANONNAME is used in different places with no good
reason. If you use the flag, the result would be in res[0].ai_canonname,
which is not used. So you can remove it and safe the elaborate resolving
which comes with it.
And I also think the comment "skip DNS loo
Rob,
Did you try to remove
hints.ai_flags = AI_CANONNAME
this flag asks getaddreinfo to return FQDN, but the function behavior is
not clearly defined for the case where FQDN is not available.
-Dmitry
On 2013-09-03 19:18, Rob McKenna wrote:
> Hi folks,
>
> Mac seems to have trouble looking up
On 09/04/2013 02:49 PM, Rob McKenna wrote:
Indeed it did.
Great, so we should behave in a similar manner to that of Apple's JDK6.
I think I understand why you added lookupIfLocalhost before getaddrinfo,
in lookupAllHostAddr, but it has the consequence of retrieving the
hostname for every oth
Indeed it did.
-Rob
On 04/09/13 10:36, Chris Hegarty wrote:
Rob,
I haven't looked at the changes yet, but I'm curious about the
decision to use getifaddrs. I know that there was a discussion started
at one point to determine what Apple's JDK6 is doing to retrieve the
local machines IP i
Rob,
I haven't looked at the changes yet, but I'm curious about the decision
to use getifaddrs. I know that there was a discussion started at one
point to determine what Apple's JDK6 is doing to retrieve the local
machines IP information. Did that discussion lead to getifaddrs?
-Chris.
On 0
Hi folks,
Mac seems to have trouble looking up local hostnames using getaddrinfo
unless a domain is set. The solution is to add a check with getifaddrs .
This fix replaces a usage of _ALLBSD_SOURCE with MACOSX. I haven't seen
a canonical answer on whether this is the way to go so I figured tr
19 matches
Mail list logo
