Zugg Software :: View topic

Posted: Wed Nov 17, 2010 5:43 pm

Hmm, that could break the other connections in the house. I'll have to do it when Chiara isn't working. I was hoping for something local to my Win7 box.

Beginner Joined: 01 Nov 2006 Posts: 27

You may also just try connecting at a peak time. Around 8PM EST the 'who' list on the MUD is larger and there tends to be more internet traffic.

I find this issue more common in the evening, which all points to your theory of breaking up network packets causing the issue.

The other thing we could try sometime if you still can't reproduce is a webx or something. However my trial period is running out so I need to decide to buy the upgrade or not soon :P

Posted: Thu Nov 18, 2010 7:24 pm

Try using NetLimiter instead. You should be able to use the trial version of Pro 3 for enough time to do a test. It alters the bandwidth for individual applications, if CMUD is limited to say 40Kbits/s it might be enough to start splitting the packets more.

Posted: Thu Nov 18, 2010 7:53 pm

The Win7 MTU route:

Run cmd as Administrator and copy and paste the following

Posted: Mon Dec 13, 2010 8:08 pm

I spent several hours on this one today and still could not reproduce any problem. Lowing the MTU on the router didn't cause any increase in packet breaks. I also tried it on the DOS command line using the netsh command. Still no effect.

So I actually modified CMUD to split the packets as they were received from the network interface. I was able to reproduce the exact Script Debugger output shown above but did not get any color problems. Nor could I get any color problems when logged into the MUD itself, even with really severe packet breaks (like every 20 characters or so).

Since I cannot reproduce the problem with the settings that you sent me on the same MUD you are using, I have no idea how what to do next with this.

Beginner Joined: 01 Nov 2006 Posts: 27

Odd that you can't reproduce.

I have since reverted to CMUD 2.37. I believe the problem still exists however it is less often a color problem, instead I will sometimes see the control characters show on the screen such as '[0;37;40m'.

Again it is very intermittent.

Given the fact that the playback doesn't reproduce the issue it seems like maybe it is a side effect of something that occurred earlier before I started recording. I'll play with the record/playback a little more and see if I can get a capture that exhibits the color issue when played back.

Beginner Joined: 01 Nov 2006 Posts: 27

So I think I'm on to something.

In order to create this trace I started the debug file before I connected to the mud, and then didn't stop until I saw the issue.

I am able to see the issue now with debug playback, on a fresh start of CMUD I do the following:

Open ansi character session offline. '#READ text.txt debug'
The second to last who command i executed improperly shows Sharaman's race as bold.

This shows in the playback every time. I have also seen the issue 'move' with some other traces where it shows the same issue the first time I play back, but then the color issue moves on subsequent playbacks.

I have sent the debug file to you in an email.

*EDIT* - I purchased 3.33a and the test was run at this version.

Posted: Mon Jan 03, 2011 6:08 pm

YES! I see the same thing when I replay your test.txt debug file here. Cool, thanks for the persistence in helping to debug this. Now that I have a test that reliably reproduces the problem, I should be able to fix it.

When looking at the debug file, the pertinent lines are:

Beginner Joined: 01 Nov 2006 Posts: 27

Hi.

Good! Progress :D

No sometimes it is the completely wrong color, I will try to get an example trace of this as well to give you more data points. If I am successful I'll send another email and post again here.

Beginner Joined: 01 Nov 2006 Posts: 27

I managed to get a trace that displays completely wrong colors as well as incorrect bold, see email for most recent traces 'text2.txt' and 'text2.raw'

Hope this helps.

Posted: Tue Jan 04, 2011 5:44 pm

Couldn't reproduce the last files that you sent, so also always try to replay the files on your end with the #READ command too before sending.

Posted: Wed Jan 05, 2011 6:21 pm

Yes! Got it this time. The difference was the Scrollback lines setting. With it set to 1000, the problem does not appear. Setting it to 10000 and using your latest test2.txt file, I was able to see the problem.

So we have a mix of two bugs here. There is the packet boundary bug with the Bold attribute not getting reset, and then there is some other bug more related to memory and scrollback lines. But at least now I can reproduce both problems and should be able to fix it for the next update.

Thanks for your patience and help with this.

Wanderer Joined: 25 Feb 2004 Posts: 98

Thanks Abascom and Zugg! I have seen this bug for years, but never went through this amount of debugging to identify what was causing it.

Posted: Thu Jan 06, 2011 1:47 am

I suggest that when you test the bugs, test them using the <dest...> feature. Scatter-writing exhibits this problem continuously. If you want to test this, just update the fields when the GMCP data comes in, you will see that eventually the colors will begin to migrate to other fields. If you like, I can send you my GMCP package and you can very this problem.

EDIT: Data sent by e-mail.

Posted: Thu Jan 27, 2011 12:00 am

Just wondering what the status on this bug is.

If you have installed and used the package I sent you the bug will be pretty obvious.

Posted: Thu Jan 27, 2011 6:36 pm

I cannot reproduce it with the stuff from Anaristos, but I did reproduce it with the stuff from Abascom. It's still on my list to be looked at for the next update. No ETA yet, sorry.