1  |   Order by index

Dec 21, 1999, 1:17am

Below is an excerpt from an email exchange with AW tech support. Here's
what I think happens. I'm calling AwTerm and AwDestroy. At one point I got
an error msg that the engine couldn't terminate the bot. I think that not
successfully terminating leaves some data structure on the AW side dirty.
This is a serious bug on their part since a connection could be lost at any
time.

I would recommend that everyone doing low level bot programming put in some
serious re-try logic for AwTerm. It may not be enough to completely fix the
problem but should help. If my theory is correct that is. It could be a
very different problem but I've satisfied myself that it's on the AW side.

Norm

______________

Well, they haven't been running. In fact I've booted my machine several
times since then. Larry is the first one that logs in so it's probably
killing the rest. Anyhow that loop only exists on your side. I just
re-booted again to make sure and got the same error.

Norm

For the account sagaman I do not see any "bot limit exceeded" errors in the
past 24 hours but it looks like the bots Larry and Moe are caught in a login
loop since 11pm PST on account i12. For some reason the bots are all trying
to login on the same socket. I suggest killing the bot program for a few
minutes and then restart them one at a time.
Please include previous dialog in all replies..
Thanks
AW Tech Support

Dec 21, 1999, 10:05am

> Below is an excerpt from an email exchange with AW tech support. Here's
> what I think happens. I'm calling AwTerm and AwDestroy. At one point I got
> an error msg that the engine couldn't terminate the bot. I think that not
> successfully terminating leaves some data structure on the AW side dirty.
> This is a serious bug on their part since a connection could be lost at any
> time.

Perhaps we should all know this and infer, but don't you call AwDestroy first
and then AwTerm?


====
Aaron Hill (Redmond, Wash.)
Electronic Mail: serac at lightmail.com
IRC Nickname: serac (on EF-Net)
ICQ UIN: 46803963
ActiveWorlds
Citizenship: serac

Dec 21, 1999, 2:00pm

You're right. But doing it backwards, or not checking for a failure of
AwDestroy can amount to the same thing.

Norm

> Perhaps we should all know this and infer, but don't you call AwDestroy
first
> and then AwTerm

Dec 21, 1999, 2:19pm

That should read "re-try logic for AwDestroy". Anyhow the idea is to NOT
shut down the SDK until you've successfully killed the bot instance. What
happens if your re-try logic still doesn't kill the bot and you eventually
have to terminate the SDK? Well, that's where the ghost bots come from.
AW isn't cleaning up their data structure. Since a lost connection, as well
as a bot programming error, can produce the same ghost, it's really up to AW
to kill the ghost.

Norm

> I would recommend that everyone doing low level bot programming put in
some
> serious re-try logic for AwTerm. It may not be enough to completely fix
the
> problem but should help. If my theory is correct that is. It could be a
> very different problem but I've satisfied myself that it's on the AW side

Mar 13, 2000, 3:41am

Hi, the only irregularity I have seen with bot limit exceeded occurs
after a universe disconnect. On reconnection, aw.dll seems to issue a
hidden aw_login which triggers the callback. At that moment, if you
hit the bot limit because there are ghost bots (not cleaned up during
the disconnection), you get this error code, which I blissfully ignore
and the bot logs back in ok.

XelaG.

[View Quote] >Below is an excerpt from an email exchange with AW tech support. Here's
>what I think happens. I'm calling AwTerm and AwDestroy. At one point I got
>an error msg that the engine couldn't terminate the bot. I think that not
>successfully terminating leaves some data structure on the AW side dirty.
>This is a serious bug on their part since a connection could be lost at any
>time.
>
>I would recommend that everyone doing low level bot programming put in some
>serious re-try logic for AwTerm. It may not be enough to completely fix the
>problem but should help. If my theory is correct that is. It could be a
>very different problem but I've satisfied myself that it's on the AW side.
>
>Norm
>
>______________
>
>Well, they haven't been running. In fact I've booted my machine several
>times since then. Larry is the first one that logs in so it's probably
>killing the rest. Anyhow that loop only exists on your side. I just
>re-booted again to make sure and got the same error.
>
>Norm
>
>For the account sagaman I do not see any "bot limit exceeded" errors in the
>past 24 hours but it looks like the bots Larry and Moe are caught in a login
>loop since 11pm PST on account i12. For some reason the bots are all trying
>to login on the same socket. I suggest killing the bot program for a few
>minutes and then restart them one at a time.
>Please include previous dialog in all replies..
>Thanks
>AW Tech Support
>
>
>

1  |   Order by index