Yeah, I think they are independent as well and the posting error isn’t consistent so hard to determine why it occurs. Increasing the timeout might not even help if its taking several minutes or if its something entirely different.
@bmnielson - I’d appreciate if you could test the changes I committed to make sure it resolves the logout issues.
I’m always using the latest TripleA prerelease triplea.jar
With the new one (released 7 hours ago) I realise that when I use TripleA to post a move I will appear as being “Online” in the “Users Online”-list of the forum ( http://www.axisandallies.org/forums/ )
and - even worse - my status is “Online” when viewing my profile summary.
I mean this all happens by only using TripleA - without any further intention to use the forum at that moment (except posting a move of course).
Personally, I would prefer that TripleA and the forum software stay independent from each other. Is see no reason for an “Online status” when I am in fact offline from the forum.
A solution (starting from the current (latest) prerelease) could be to add an option such as “Post and stay logged in after posting” AND “Post and log out from the forum after posting”.
Yeah, I think they are independent as well and the posting error isn’t consistent so hard to determine why it occurs. Increasing the timeout might not even help if its taking several minutes or if its something entirely different.
It is definitely independent. I noticed - when trying to post a move - that TripleA logged me in to the forum (I appeared as “Online”) but the error message (Timeout) occured nevertheless.
@P@nther - This is a somewhat different issue that is really a forum bug. The ‘users online’ list appears to check if you have any active sessions not if you have an active session for the given browser. So this issue isn’t just a TripleA issue and can be reproduced by doing the following:
1. Open Chrome and login to axisandallies.org
2. Open Firefox and navigate to axisandallies.org and you’ll see that it shows you in the ‘users online’ but you can’t post since you aren’t logged in on Firefox
Sure, I have understood the issue and fully agree. I see that this is a forum software issue.
I just wanted to point out the downside of the new (triplea.jar) modification:
An unintended “online-status” when there was no intention for activity on the forum except posting the move through TripleA.
Fair enough. Though I’m guessing the vast majority of users stay logged in forever in the browser so won’t even notice. An option could be added as you described to allow the user to choose whether to login or logout after TripleA posts but I honestly don’t think hardly anyone would ever choose to logout.
I agree. If it really bothers you, you can choose to use a separate account for posting from triplea, so at least the forum account won’t show up.
Fair enough. Though I’m guessing the vast majority of users stay logged in forever in the browser so won’t even notice. An option could be added as you described to allow the user to choose whether to login or logout after TripleA posts but I honestly don’t think hardly anyone would ever choose to logout.
Maybe. And not everyone would probably realize that they were logged (and kept in) to the forum by TripleA.
I don’t have a fix on the forum side. Hopefully a fix on the TripleA side will work.
Why is it not possible to fix the forum side?
The bug isn’t really triplea related, since you can easily reproduce the bug using just 2 browsers….
Not to be glib but if you want to dig through the spaghetti code that is SMF and build a mod (as in modification no module) that won’t break everything else, please do so.
I’ve dug into the innards of SMF before spent hours on issues only to get basically nowhere, realizing that if I touch “that code right there” to fix the issue, everything else come crumbling down.
What version of SMF was running on the old site? I can try doing a quick diff to see if anything’s changed in the logout module.
Realized just after my post that I could just check an archive.org snapshot of the forum. Looks like it went from 1.1.16 to 1.1.21. To spaghetti land I go…
Luckily not much was changed so it was pretty easy to track it down.
In 1.1.19 they added some code in LogInOut.php that changes the users password salt whenever you log out. When you switch to another browser, the hash matching fails since it was generated with a different salt, and you get logged out.
The offending code is at lines 463-464. I have no idea how to make a modification for SMF but I guess you could just try commenting out those lines to start with, to see if it works.
The call to session_destroy() on line 462 is also added in that change, but I would expect that is independent for each browser session so I don’t think that’s causing any issues.
bmnielsen, you rock!
I just get angry when I look at SMF code. :wink:
I’ll see what I can do but it seems like this might be a security fix/hack.
Yeah, I thought of that as well. The changelog just says “Sessions weren’t always cleaned up properly on logout” though, which doesn’t really indicate a security issue. I’m thinking that adding the session_destroy was the important thing for that.
Update? Were you able to fix everything on the forum side? Should I revert the changes I made to TripleA to remove the logout step?
No. I’m not going to have time to touch anything until after GenCon.
Alright, well we are planning to release the next stable version of TripleA shortly. Did you want me to release the change to remove axisandallies.org session logout when posting games?
Alright, well we are planning to release the next stable version of TripleA shortly. Did you want me to release the change to remove axisandallies.org session logout when posting games?
Yes, let’s try that.
I got around to posting a failing turn summary in debug mode. The forum is just returning 404 not found with no additional information. I believe this is in line with what someone else reported earlier when they were manually attaching a file.
So not really anything to do about it in TripleA, besides potentially showing a different message when that happens.
