RE: [Quickfix-developers] [qfj] connect and logon in wrong order after disconnect?

[Steve B. <st...@te...>]

>Graham Miller wrote:
>I was going through the certification process with Archipelago using =
qfj.=A0
I have managed >to pass all the tests except the last one, which =
requires
that your engine recover >gracefully from an unexpected disconnect.=A0 =
It
appears that perhaps after an unexpected >disconnect, my engine is =
trying to
send a logon message before the socket is >reconnected.=A0 At least =
that's
what the log would indicate:

>...

>I looked through the code a little bit and it seems like there is no
mechanism to ensure >that the logon is after the socket reconnect, but =
I'm
pretty unfamiliar with it, so I >wouldn't be surprised if I'm missing
something.
>
>I am using qfj beta 1, and have seen this behavior on both a windows =
and a
linux machine.

Hi Graham,

I think I've located the problem. Are you compiling QFJ from the beta
sources
or using the JAR file? If you are compiling the sources, I can send the
patch
I've made to the QFJ CVS branch (or I could give you the version number =
of
the updated file you could get it from CVS).

I'm leaving on an international trip this afternoon so I don't have time =
to
do a full beta release with the bug fix. However, if you aren't =
compiling
from
the sources I can build a JAR file with the fix and email it to you if =
you
let me know within the next few hours.

Thanks for reporting this issue. The tests for the initiators aren't as
comprehensive as for the acceptors. I've added an automated regression =
test
for this issue.

Steve