Hi,
I am sorry, but I do not know how the perl-ldap maintainers shall react to
this bug report.
With the exception of some symbolic constants from Socket.pm, perl-ldap
only uses Socket.pm indirectly via the IO::Socket::* family.
So, I fear there is nothing perl-ldap can do here to fix the issue-
Would you mind to inform the Socket.pm author about your finding?
Because I consider it a bug in a module if some of its - directo or indirect -
clients start to fail after a new release has come out.
But I admit I have doubts that Socket.pm changed in an incompatible way.
The only effective change between Socket 2.006 and 2.007 is:
--- Socket-2.006/Socket.xs 2012-08-19 23:31:04.000000000 +0200
+++ Socket-2.007/Socket.xs 2012-12-16 19:24:00.000000000 +0100
@@ -916,13 +916,13 @@
case AF_INET:
if(addrlen != 4)
croak("Bad address length for Socket::inet_ntop on AF_INET;"
- " got %d, should be 4");
+ " got %d, should be 4", addrlen);
break;
#ifdef AF_INET6
case AF_INET6:
if(addrlen != 16)
croak("Bad address length for Socket::inet_ntop on AF_INET6;"
- " got %d, should be 16");
+ " got %d, should be 16", addrlen);
which
a) fixes a real bug: add argument for the %d pattern
b) does not look as if it would suddenly break working implementations
of perl-ldap only.
In addition to that, I cannot reproduce the issue with Strawberry Perl
5.16.2.1 (with Socket 2.006) when connecting to a non-AD LDAP server.
So:
* Can you try reproding the issue using another LDAP server (preferably
another type. i.e. no AD)?
* Can you try reproducing the issue using the above version of Strawberry Perl
(
http://strawberryperl.com/)?
* Can you double check your implementation?
I assume you did it for yourself, and did not use pre-compiled images from
somewhere on the net?
Please report back.
Thanks
PEter
--
Peter Marschall
pe...@adpm.de