Zugg Software :: View topic

Posted: Sat May 24, 2008 2:24 pm

The old thread was getting confusing and it already had some of its original issues fixed. I've started this new thread for 2.25 onward ATCP bleedthrough issues. I'm still seeing a lot of them, albeit less than before.

Zugg: This was repeatable and I even turned off triggers to capture this. I do not this this will solve all the bleed through issues, but perhaps a portion of them.

What I see: (Error in bold)
7700h, 8107m, 33900e, 23419w cexdb-|0h(0%):+12m(+0%)|ql
Entering the Main Gate of Ashtan.
The city gates of Ashtan stand proudly here, black and gold banners waving defiantly at those who would challenge the Bastion of the North. A runic totem is planted solidly in the ground. A cute little humgii is sitting here placidly, attached to a leash. A gleaming wooden sign is posted here. Overflowing with blood and other trophies of the dead, a sacrificial altar to the Lord of Warfare has been erected nearby. Polished marble gleaming, the Effigy of Victory resides here, a testament to the favour of the Great Mother. There are 2 Mhojave falcons here. Ever-vigilant, a Zarathustran eagle monitors the area for threats. A Rajamalan woman has been nailed upon a ridged, bloodstained cross and left to perish. A sigil in the shape of a small, rectangular monolith is on the ground. Wings flaring, a pegasus tosses his regal head impatiently. A canvas backpack lies face-down here. Scuffing its hooves on the ground nervously, a striped zebra glances at you. There are 8 Shadow Amazons here. There are 10 War Witches here. The bloodied body of an innocent little girl is here. A black marble bench is here.
You see exits leading.Exits n,ne,s,nw north, northeast, south, and northwest.

#debugfile Output

Posted: Sat May 24, 2008 3:58 pm

This is probably a different mechanism from the one I posted above. This one has a ton of trigger activity going on and I've captured the debugger window output along with the #debugfile output and, of course, what I actually see.

Here's what I see. I've bolded the erroneous parts.

7700h, 8095m, 33786e, 27750w cexkd-|+35h(+0%):-12m(0%)|ii corpse
more
more
more
7700h, 8107m, 33846e, 27822w cexkd-generosity

UNDEF selfishness
7700h, 8105m, 33846e, 27821w cxkd-|0h(0%):-2m(0%)| (Not Selfish)
Offering Corpses: 55
selfishness
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w00 M:8105/8107 E:33846/33900 W:27821/32400 NL:0/100 cexkd-
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w[Acropolis]: Failsafe: eq
0 M:8105/8107 E:33846/33900 W:27821/32400 NL:0/100 cexkd-
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)
You rub your hands together greedily.
7700h, 8105m, 33846e, 27821w cxkd- (Not Selfish)

Offered Summary 3:
29 a fairy Knight of Sidhe
21 a fairy Lady of Sidhe
5 a cunning black wolf

7700h, 8105m, 33846e, 27821w cexkd- (Not Selfish)

#debugfile

Posted: Sat May 24, 2008 4:02 pm

Something happened in the previous post with the debugger window. Trying again here:

Posted: Mon Jun 02, 2008 7:04 am

With the 2.26 version around the corner, any hope we can get some ATCP bleed-through fixes in as well? Now that I've been playing with 2.25 a while I'm seeing this bleed-through it all over the place.

Posted: Mon Jun 02, 2008 5:25 pm

I am still having a hard time reproducing this. It doesn't actually look like a real packet-boundary problem anymore. You have some pretty complex scripts running there and I think one of them is causing some output that happens in the middle of an ATCP command that might be causing the problem.

Can you show me the full Telnet trigger that you have running for the Char.Vitals? Also, do you have any Expression triggers that might fire when the Char.Vitals trigger is running?

Posted: Mon Jun 02, 2008 11:07 pm

Posted: Mon Jun 02, 2008 11:11 pm

I'm willing to put forth some effort here to track down whatever is causing this. If you can think of any other debugging techniques I could employ (and that would be helpful to you) please let me know.

Perhaps I could come up with a simple package I could send you that exhibits this issue. Would that be helpful?

Posted: Tue Jun 03, 2008 4:41 pm

Yes, if you had a package and instructions for me to reproduce this on a specific MUD, that would help a lot.

It's possible that ATCP triggers are not disabled by the global trigger flag. I need to look into that, but ATCP triggers are pretty new and are fired in a low-level part of CMUD, so they might be running. You can probably verify that easier than me right now, but that's an idea I had in the shower this morning. So you might need to disable the ATCP trigger manually (specifically disable it in the settings editor).

But any way you can come up with something I can use to reproduce this would be great. It's something more complex than just packet boundaries. Maybe it's the interaction with the mapper via the #OK or #TAG command that is doing something. Or maybe it's something in one of the Events that you are raising in the ATCP trigger.

Posted: Wed Jun 04, 2008 2:04 am

Perhaps you could provide me with another tool or two in my efforts to replicate this for you.

Is there any way I could replicate the server sending me ATCP lines? It would be extremely helpful if I could simulate the sending of the server codes so that I could do this off-line and not at the mercy of the server.

I often make heavy use of #show when testing triggers and debugging issues. At this time I know of no way to simulate the reception of #telnet 200 type lines.

Posted: Wed Jun 04, 2008 5:21 pm

Yes, the #SHOW command will do this. The #SHOW commands stuffs the data into the network buffer in CMUD to simulate incoming data from the MUD. When doing this, you actually use the #SHOWPROMPT command (since the #SHOW command adds a CR/LF that you might not want).

For example:

Posted: Thu Jun 05, 2008 12:38 am

Thanks Zugg, this should help a lot.

I have a few clarifying questions:

1) What in the #showprompt line above makes it a telnet 200 type line? I don't see anything in particular about the line that makes it different from another.

2) Can you replay the human readable debugfile or only the raw one? Since the human readable one is a lot easier to work with it seems like it would be the more natural choice to use.

Posted: Thu Jun 05, 2008 9:05 am

The %char(255)%char(250)%char(200)...%char(255)%char(240) part is the part that makes this line contain suboption information. A line isn't a "telnet 200" line, the suboption text is recieved as part of an otherwise normal line - in this case, your prompt. The suboption text is delimited by IAC SB OptionNum Text IAC SE where everything except text is a single character - the %char commands above create the right characters. OptionNum is the ASCII character that corresponds to the number of the suboption - ATCP's number is 200, so that makes this ATCP text. You can read about telnet suboptions and how they work in RFC 855.

For your second question, no. That's why it records two files - one for replaying, and one for reading. If it could replay the readable file, the replayable one would be pointless.

Posted: Thu Jun 05, 2008 9:09 pm

Ok, I verified that turning off triggers will also turn off Telnet triggers, so that's ok.

What I just need you to do then is to post the Script Debugger output (with the Raw Input/Output message enabled) and post the results again, but with your triggers turned off. You said that it still happens when triggers are turned off, so posting that example should be a lot easier to figure out since it won't be full of your other triggers all running.

Posted: Fri Jun 06, 2008 12:40 am

Okay, I plan on doing some detailed debugging on bleedthrough this weekend. I think I have all the tools I'll need to get some results.

Posted: Sat Jun 07, 2008 3:30 pm

I logged in on a clean/new session that has absolutely 0 triggers. It has ATCP enabled and compression disabled.

I found one room that was consistently giving errors and recorded what I saw along with #debugfile and debugger results. I noticed that if the room description was changed (by having more or less stuff in there) the error I was seeing changed as well. So I included two variations below, the only difference is the length of the room description.

What I saw:

Central Market of Ashtan.
A sewer grate looms darkly beneath your feet. A cute little humgii is sitting here placidly, attached to a leash. A runic totem is planted solidly in the ground. Leaning against the wall, a Soldier of the Golden Apple Corps watches the goings on of everything around him. There are 7 Knights of the Illuminati here. A gleaming wooden sign is posted here. There are 6 War Witches here. Seething with primal energy, the Well of Chaos stands here firm, a circle of divine power. Sprightly balancing on his hands, an Ashtan Joker cavorts here in a playful manner. A Zarathustran eagle perches atop a monolith, a golden plaque clutched within its talons. Brandishing a razor sharp blade, a Freedom Fighter stands here, twirling his long mustachios. Wings flaring, a pegasus tosses his regal head impatiently. You see a sign here instructing you that WARES is the command to see what is for sale.
You see exits leading north, northeast, east, southeast, south, southwest (open door), west, and northwest.
1634h, 1318m, 7066e, 4895w ex-
al Market of Ashtan

Posted: Mon Jun 09, 2008 4:52 pm

I got your file and instructions on this. I'll see if my character on Achaea is still active and if I can get to that location (I have no idea what town I was in when I last saved).

Posted: Mon Jun 09, 2008 11:27 pm

If you are below level 10 just type 'portals' then follow the signs to Ashtan 'read sign'. Once you are there you can use the map and directions I provided. You could also use the 'newbie' channel to get help. I.E. 'newbie how do I get to Ashtan's main gate?'. Failing that you could probably just ask anyone you see and they'd probably be able to help you.

At this time the room I pointed out is consistently showing that error but it isn't always that way. I think it has to do with the length of the room line, so if it some point in the future the amount of junk in the room changes that room may no longer show the issue and I'll need to find you another room. I'll try to keep an eye on it for the next little while and try to find you another room if something happens to the room I pointed out.

Posted: Tue Jun 10, 2008 4:37 am

You might also want to post an actual screenshot image so that I can try and mimic the same client width as your window if the length of the room line makes a difference. But I'll try logging into Achaea tomorrow and let you know if I see it.

Posted: Tue Jun 10, 2008 4:48 am

I had already sent you a screenshot in the e-mail from the weekend. I included two pictures in that e-mail, one of the results of doing a 'ql' with the error clearly visible and the other a map to show you how to get there.

Screen size isn't a factor as it happens both on my main package and on a blank session which has a totally different screen width.

Posted: Tue Jun 10, 2008 2:10 pm

Blasted, I check the room just now and someone modified the contents and it no longer shows the bleed-through. I'll post again when it changes back and is visible again.

Posted: Tue Jun 10, 2008 2:34 pm

Okay, I figured out a fool-proof alternative method for you to see this. I was ruminating about repeatable ways to show this when it hit me. I have long suspected this has something to do with the size of the data chunk being sent by the mud and I connected that just now with my messages showing this at times. It turns out that if I have just the right amount of test in my messages I'll see this as well. So I hopped on a character in a new session and I gave it a test.

This is repeatable and I have no triggers in this session.

Procedure:

Posted: Tue Jun 10, 2008 9:41 pm

When I tried that, I only get the first 20 messages shows. This is with a new character. So there is probably some setting that I need to give the MUD to tell it to display more than 20 messages in a page. let me know how to do that.

Edited: Nevermind, I found the CONFIG command. So I did a CONFIG PAGELENGTH 40 and then I did a MSGS to display all 27 messages. Unfortunately, I still didn't get any bleedthrough problem.

It's possible that there is a packet size setting in Windows or in your network software, or in your router that might be changing this too. Have you made any changes to your network setup at all?

Is there *anyone* else playing Achaea that can reproduce this problem also? It's driving me crazy and without some more help on this, it's not going to get fixed in 2.27.

Posted: Tue Jun 10, 2008 11:44 pm

Argh, this is so vexing. I found one other config option that might be causing you to not see it.

Try setting your prompt to show endurance and willpower as well as I think it defaults to only health and mana. When I tried it just now I didn't see the bleedthrough occur when only health and mana where shown.

Type:
config prompt all

Here are my other settings:

Posted: Tue Jun 10, 2008 11:57 pm

Zugg. I want to suggest a slight alteration of the procedure I suggested above. Since this seems dependent on the total quantity of text sent, I increased the length of the message I was sending myself and then I started at 0 and started sending myself the longer message. I checked 'msgs' each time I sent a message and with the longer message I got the error at 14 messages.

So perhaps try this modified version:

Posted: Tue Jun 10, 2008 11:58 pm

config clothesline? :S

Also, I like the way you say "I'm temporarily in Malaysia" like you fell asleep on the train and woke up there by accident. Blast!