quickfix-developers

[Quickfix-developers] Proper way to disconnect session

[Eric G. <er...@la...>]

I'm having trouble with reconnecting after I disconnect a session.  I'm =
using Session::logout() to kill the session and I see the logout message =
coming back from the targetcomp. The initiator thread that is supposed =
to terminate at this point is hanging somewhere and when I try to =
reconnect, it fails because the socket initiator is still waiting for =
the variable m_stop to turn to true.

Does anyone else have this problem or is there a better way to =
disconnect?

Thanks,

Eric Ginder
Last Atlantis Capital

Re: [Quickfix-developers] Proper way to disconnect session

[Oren M. <or...@qu...>]

Why is the initiator thread being terminated?  All you need to do is=20
call logout when you want to disconnect, and call logon when you want=20
to reconnect.  Nothing here should affect your Initiator thread.  Is=20
something else going on?  Did you call stop on the Initiator?

--oren

On Jul 7, 2004, at 4:44 PM, Eric Ginder wrote:

> I'm having trouble with reconnecting after I disconnect a session.=A0=20=

> I'm using Session::logout() to kill the session and I see the logout=20=

> message coming back from the targetcomp. The initiator thread that is=20=

> supposed to terminate at this point is hanging somewhere and when I=20
> try to reconnect, it fails because the socket initiator is still=20
> waiting for the variable m_stop to turn to true.
> =A0
> Does anyone else have this problem or is there a better way to=20
> disconnect?
> =A0
> Thanks,
> =A0
> Eric Ginder
> Last Atlantis Capital
> =A0

Re: [Quickfix-developers] Proper way to disconnect session

[Eric G. <er...@la...>]

Oren,

Thanks for the hint.=20
I was calling pInitiator->start(); but not using logon(). Somehow the =
session logs on automatically when you do this, and it never occurred to =
me that logon() was missing.=20

When I added the logon() statement, the session reconnects just fine.=20

What this means however, is that the threads on the initiator or =
acceptor never actually get terminated as long as your application is =
running. Is that correct?

Eric

  ----- Original Message -----=20
  From: Oren Miller=20
  To: Eric Ginder=20
  Cc: qui...@li...=20
  Sent: Wednesday, July 07, 2004 6:09 PM
  Subject: Re: [Quickfix-developers] Proper way to disconnect session


  Why is the initiator thread being terminated? All you need to do is =
call logout when you want to disconnect, and call logon when you want to =
reconnect. Nothing here should affect your Initiator thread. Is =
something else going on? Did you call stop on the Initiator?

  --oren

  On Jul 7, 2004, at 4:44 PM, Eric Ginder wrote:


    I'm having trouble with reconnecting after I disconnect a session.  =
I'm using Session::logout() to kill the session and I see the logout =
message coming back from the targetcomp. The initiator thread that is =
supposed to terminate at this point is hanging somewhere and when I try =
to reconnect, it fails because the socket initiator is still waiting for =
the variable m_stop to turn to true.
    =20
    Does anyone else have this problem or is there a better way to =
disconnect?
    =20
    Thanks,
    =20
    Eric Ginder
    Last Atlantis Capital
    =20

Re: [Quickfix-developers] Proper way to disconnect session

[Oren M. <or...@qu...>]

Yeah, though they won't really be doing much.  They will essentially be=20=

blocked on a select call, waking up every second or so to check if they=20=

should log on again.  If you are using a SocketInitiator or Acceptor,=20
all your sessions are sharing one thread anyway.

--oren

On Jul 8, 2004, at 12:55 PM, Eric Ginder wrote:

> Oren,
> =A0
> Thanks for the hint.
>  I was calling pInitiator->start(); but not using logon(). Somehow the=20=

> session logs on automatically when you do this, and it never occurred=20=

> to me that logon() was missing.
>  =A0
> When I added the logon() statement, the session reconnects just fine.
>  =A0
> What this means however, is that the threads on the initiator or=20
> acceptor never actually get terminated as long as your application is=20=

> running. Is that correct?
> =A0
> Eric
> =A0
> ----- Original Message -----
>  From: Oren Miller
> To: Eric Ginder
> Cc: qui...@li...
> Sent: Wednesday, July 07, 2004 6:09 PM
> Subject: Re: [Quickfix-developers] Proper way to disconnect session
>
> Why is the initiator thread being terminated? All you need to do is=20
> call logout when you want to disconnect, and call logon when you want=20=

> to reconnect. Nothing here should affect your Initiator thread. Is=20
> something else going on? Did you call stop on the Initiator?
>
> --oren
>
> On Jul 7, 2004, at 4:44 PM, Eric Ginder wrote:
>
>
> I'm having trouble with reconnecting after I disconnect a session.=A0=20=

> I'm using Session::logout() to kill the session and I see the logout=20=

> message coming back from the targetcomp. The initiator thread that is=20=

> supposed to terminate at this point is hanging somewhere and when I=20
> try to reconnect, it fails because the socket initiator is still=20
> waiting for the variable m_stop to turn to true.
> =A0
> Does anyone else have this problem or is there a better way to=20
> disconnect?
> =A0
> Thanks,
> =A0
> Eric Ginder
> Last Atlantis Capital
> =A0