Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

Subject: Re: [OMPI devel] Libevent visibility problem
From: Barrett, Brian W (bwbarre_at_[hidden])
Date: 2011-07-12 18:30:27

On 7/12/11 4:21 PM, "Ralph Castain" <rhc_at_[hidden]> wrote:

>On Jul 12, 2011, at 12:29 PM, Barrett, Brian W wrote:
>> On 7/11/11 4:31 PM, "Ralph Castain" <rhc_at_[hidden]> wrote:
>>> On Jul 11, 2011, at 2:51 PM, Barrett, Brian W wrote:
>>>> Hi all -
>>>> When libevent was made its own component last fall, it appears that
>>>> function renames and visibility settings were lost. This is proving
>>>> rather problematic for a project I'm trying to get running with the
>>>> trunk
>>>> which uses libev (which provides a libevent compatibility layer). It
>>>> works fine with Open MPI 1.4 and 1.5, which rename the libevent
>>>> functions
>>>> in rename.h, but causes symbol clashes for the trunk. This seems
>>>>like a
>>>> rather glaring oversight, especially since there was a bunch of work
>>>> done
>>>> to remove the visibility flags from the libevent component.
>>>> At one point, someone was running a symbol checker to make sure things
>>>> like this didn't happen; has that been killed off?
>>>> Can whomever is responsible for the new libevent please fix this?
>>> I guess that would mostly be me - my penalty for doing the update. Can
>>> you tell me which symbols are clashing? We honestly thought we had that
>>> all worked out, but I don't know of any tool for checking it other than
>>> the basic ones.
>> It actually looks like all the event symbols are public in the trunk
>> (1.4/1.5 are fine). Most of the symbols will have to be public, but
>> renamed (from event_add to opal_event_add, for example) in the previous
>> releases. That seems to be the missing piece.
>I see the problem - I'll try to fix it shortly. Might take me a few days
>as I'm in the middle of some other stuff.
>Sorry about that - we thought we had it, but missed. :-(

No problem, I figured out a temporary work-around on my end, so I'm not in
a huge rush.


  Brian W. Barrett
  Dept. 1423: Scalable System Software
  Sandia National Laboratories