Jump to content
Search In
  • More options...
Find results that contain...
Find results in...
VinceDSS

Final DOOM demos [-complevel 4]

Recommended Posts

On 2/10/2019 at 12:02 AM, Grazza said:

Chocolate Doom (at least, the old version that I tested it with).

  Reveal hidden contents

T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55376, sector: 114, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55376, sector: 115, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55376, sector: 116, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55377, sector: 114, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55377, sector: 115, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55377, sector: 116, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55378, sector: 114, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55378, sector: 115, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55378, sector: 116, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55379, sector: 114, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55379, sector: 115, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55379, sector: 116, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55380, sector: 114, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55380, sector: 115, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55380, sector: 116, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55381, sector: 114, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55381, sector: 115, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55381, sector: 116, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55382, sector: 114, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55382, sector: 115, complevel: 4
T_MovePlane: Stairs which can potentially crush may lead to desynch in compatibility mode. gametic: 55382, sector: 116, complevel: 4

 

I'd be curious to know which Choco version that was, because someone *ahem* changed the code to initialize the 'crush' field as well:

https://github.com/chocolate-doom/chocolate-doom/commit/bfda08cf5275a70a705b3d290ca5271d40481993

 

Discussion leading to this commit;

https://github.com/chocolate-doom/chocolate-doom/issues/368#issuecomment-76356171

Share this post


Link to post

It was an ancient version (dated 13-02-2006) that happened to be the one in my Doom2 directory.

 

Ah, so it seems that it is no longer true that Choc (and Crispy) do "not touch the memory manager and the sizes of all internal data" and therefore cannot retain general compatibility with vanilla in these cases. I didn't realize that had changed.

 

Andrey chose values that fixed the cases he knew about at the time, but understood that it would not work in all cases (it happened to work for mm09-512.lmp too, as it turned out).

 

So it seems that now none of these ports can reliably emulate the vanilla behaviour, but have different incompatibilities. That's a shame. If that can't be remedied, it would be nice if they could at least be consistent with each other.

 

Incidentally, Dime seems to have been very unlucky in this respect. No other demos that I am aware of have desynched due to this happening on ev21 (single level or movies), and it has happened to him twice.

 

Just to be clear: the stdout output that I quoted was from current Prb+.

Edited by Grazza

Share this post


Link to post

Ty Andrea!

 

My stream was dropping frames leading into the final maps of evilution and I commented to a chatter "no worries Im recording a demo so the final product will be untarnished". Excitedly I upload it and zeromaster comments in a speedrunning discord that the desync happened again in map 25. From elation to deflation. 

 

Back in 2014 I swapped from chocolate doom to cndoom because map 30 of plutonia is prone to crashing if you use the BFG while the redrock wall lowers to the icon of sin but cndoom and crispy brought in desyncs on administration. 

Share this post


Link to post

Clearly the only solution here is to run 30ev in dosbox :)

 

Also this was actually the 3rd time it happened to dime, I never had it happen myself. I think if anyone is going to do 30ev again it's probably best to activate the stairs, then kill the imps, as this should gurantee no desync happens, at least in this case.

 

I am just so happy that my nightmare run didn't desync to this.

Share this post


Link to post

iirc there was the same issue with my old run in prb+ desyncing in vanilla

Share this post


Link to post
19 hours ago, Grazza said:

Ah, so it seems that it is no longer true that Choc (and Crispy) do "not touch the memory manager and the sizes of all internal data" and therefore cannot retain general compatibility with vanilla in these cases. I didn't realize that had changed.

 

Leaving this value in the struct uninitialized doesn't mean it gets the same random value as Vanilla would - that's impossible to achieve. Thus, we decided to initialize it with *some* value, but apparently the one I chose introduced bias. I am very sorry for this and apologize for my mistake! A PR to help mitigating the issue has already been filed: 

https://github.com/chocolate-doom/chocolate-doom/pull/1137

 

Share this post


Link to post

Nice. BTW, there is a mistake in the txt in the times for maps 15, 31 and 32. (I haven't checked them all, but these stood out.)

Share this post


Link to post

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×