Demo combiner

[Ostap]

I mean, after I upgraded Fix42Mod to latest version, and then combined the demo, I didn't lose any part. The demo worked 100% with no desyncing.

Besides - if you take a closer look at the .bst file - you'll see there is a client ID - all bad syncs are per player !
Bad syncs present in demo file - not necessary break the whole demo ..... have you checked all players ?

[Marco]

It shouldn't .... that demo has no "bad-sync" times recorded .... frankly - i'm not sure how it got better


Demo is not an archive ....its a binary file with bunch of data dumped into it ....(world state and all the player/monster actions that were present during the game)
If you know the structure of file - its not a problem to read it (structure of the file you can derive from C++ code in distributed 1.10 SDK of the game)

So: I can learn about ".dem" structure on 1.10 SDK, but there isn't a program that opens them... Do I have to compile it in C++, based on ".dem" structure? Correct?

Besides - if you take a closer look at the .bst file - you'll see there is a client ID - all bad syncs are per player !
Bad syncs present in demo file - not necessary break the whole demo ..... have you checked all players ?

No, it didn't give me any bst for this demo. I didn't combine it previously (2+ weeks ago, when I didn't update Fix42Mod yet), because I manually checked all parts before.

I have to check other demos and see if there are any broken. Eventually, would be Notepad++ enough for opening bst files?

[Ostap]

So: I can learn about ".dem" structure on 1.10 SDK, but there isn't a program that opens them... Do I have to compile it in C++, based on ".dem" structure? Correct?

Yes but why ? ...unless you planning to contribute to the project ...

Eventually, would be Notepad++ enough for opening bst files?

Yep..

Also once you'll get the .bst file .... - you can control when demo will be cut ...
(open for editing the ".bat" file ....you'll see the the command: "tools\42DemoFileCombiner.exe Combined_Demo 1")
That first parameter is the filtering value for bad-syncs ...(default =1 .... cut at first bad sync)

Attachment 6610

In the .bst file - you can see that player had bad syncs ...but recovered a few times....1,2...RECOVERED.... 1,2...RECOVERED
So ....you can play with the parameter ....and set it to ...for example 14 ...(after 14 counted bad syncs demo will be cut)
This will allow to get bit more of possibly OK frames into working demo ...or even skip the demo splitting ...if there were just few bad syncs. (Max filtering value is 1200 = 60 sec)

[Ostap]

Minor update on tweaking the "cut off" parameter:


After checking some demos - If I correctly understand the way it all works - then amount of bad syncs is VERY relative number
It all depends if player had any interactions during bad sync time

If he was just running around without shooting anyone or being shut by anyone - huge possibility that even at high count of bad syncs - demo will still be OK !

[Marco]

There is a little but
Demos generated on 42 prior 30 Nov 2018 ...are a bit out of sync ...each part is missing few frames...
so when combining old demos - you'll get a few messages in the game log - about sync ...

I want to UP this because I did the experiment last september. Well, I saved a lot of 42demos since 2013.

I gave the combiner a try with a demo from september, but 2018 (Sep-29-2018, a high score of BadBoy by 34-0*, in which I was part of; he played under the alias of BUBLIK; Map: The Lost Tomb, on Frag 07).

Basically the combiner does the job, but when playing the demo, it just freezes when it reaches the, let's say, point-break of the first part. For instance: if a random demo's Part #1 ends at 3:01, the combined old demo will freeze at that point.

So, my question is: how did you find out that demos prior Nov-30-2018 don't work, while after that date they do? Is the combiner coded specifically for newest demos? Which are the parameters that caused this?

Note: I took the combiner exe and files needed from a random demo (a recent one, from 2020) and then I started my test.

*Indeed it was actually 33-1, since I managed to throw him into the trap!

[Ostap]

So, my question is: how did you find out that demos prior Nov-30-2018 don't work, while after that date they do? Is the combiner coded specifically for newest demos? Which are the parameters that caused this?

Coz on Nov 30 i've made the fix ....that will stop and restart demo within 1 tick ....without loosing anything
all demos prior to that date - were restarted using old command ...and would loose about 1 sec of data ....

[Marco]

Coz on Nov 30 i've made the fix ....that will stop and restart demo within 1 tick ....without loosing anything
all demos prior to that date - were restarted using old command ...and would loose about 1 sec of data ....

You mean an internal command in the combiner?

Because I've tried to unfreeze the demo with "dem" commands in Sam, but demo still remains frozen at its part checkpoint.
And it will terminate execution within 1-2 mins...

[Ostap]

You mean an internal command in the combiner?

Because I've tried to unfreeze the demo with "dem" commands in Sam, but demo still remains frozen at its part checkpoint.
And it will terminate execution within 1-2 mins...

Nop ... within the server it self
You should play demos apart and then combine recorded videos afterwards...