The Next Iteration of DSDA

[GoneAway]

Shock's run is crosslisted there. It's a pacifist run that is also appearing in the uv speed area for convenience. Probably that's just injecting the run at the top of the cell and doesn't bother sorting anything. That's a small bug I guess.

[vdgg]

10 Sectors

MAP13 has two runs marked as records. Weird. If you go to Daiyu Xiaoxiang page, record view, this 4:24 doesn't appear (which is correct).

MAP30, the same, except Yashar's demo should not only have the record status removed, it should also be moved to Other (UV Max attempt. Misses one monster).

 

Community Chest

MAP20 blob's demo misses many monsters, so not UV Max

 

eob11

E2M9 Max by me has incorrect time

 

Eternal Doom

MAP28, my TAS is also reality

 

FCFF

MAP05 UV Max is also reality

 

Marsw301

MAP21 Max by me has incorrect time

MAP29 Max by me (the slower one) has incorrect time

 

Mini Level Megawad

MAP03 Phml's demo misses two monsters

 

Reverie

MAP29 danne's Max should be a record, not Rizera's. The old version has no timed pillars lowering, so it is much easier to get a better time.

 

 

Edited May 24, 2020 by vdgg

[baja blast rd.]

7 hours ago, vdgg said:

Reverie

MAP29 danne's Max should be a record, not Rizera's. The old version has no timed pillars lowering, so it is much easier to get a better time.  

 

 

18 of pwfinal is another one I remember. I think it makes sense for a 'plays back with [different wad]' note to preclude the record icon. 

[GoneAway]

Usually double record icons are coming from an issue that was fixed some time ago but where the cached table entry hasn't been refreshed yet. As far as "plays back with old version" demos go, these will probably get removed eventually unless they're historically relevant, in which case they'll go into Other. If there's a high enough volume then a separate wad page should be made.

[vdgg]

I was wondering, why are two runs still on the old version of Scythe2 (MAP26)? I didn't officially report it, but I thought @kraflab acknowledged that when I mentioned Aquasa's run in Notable 2019 demos thread.

 

Here's a controversial one: Method's UV Speed on MAP30 should be moved from marsw301 to marswar.

The date is 2003 so it was obviously recorded with the old version.

I think it is in the new version on the assuption that it does not go out of sync when playing back with it.

But it does descync, only slightly, it still manages to produce map exit, but the time is different: 0,34.63 with marswar.wad ; 0,34.66 with marsw301.wad.

[GoneAway]

I'm doing some framework upgrades right now.

 

If you run into any issues / error messages / weirdness in the next days please let me know.

 

vdgg: not sure why I didn't have a note about that aquasa demo but it's on the list now. I agree the marswar demo can be moved as well.

[plums]

Hey, I'm not sure where the best place to put this is or anything, but: the demo in this youtube video desyncs.

Final Doom TNT Evilution Map29 UV Speed in 1m48s by Method

https://www.youtube.com/watch?v=_5Ohu4SlSAo&list=PLTmV4zpWzC8lK_3jZ8qeTwfYb2ugKkvnR&index=33

 

[Grazza]

This is a -complevel 2 demo and it plays back fine. If you let prb+ use its autoloading patterns, then you don't have to do anything special to watch demos like this.

 

If Compet-n Final Doom demos are instead made into videos (or played back) with -complevel 4 forced unconditionally, then about 30% of them will desync (a higher percentage for longer ones, lower for shorter ones). Makes me wonder why I ever bothered with the autoloading patterns, which should have confined most issues of this type to the dumpster of demo history - seamlessly and without user intervention required.

 

BTW, the text-files are not a reliable guide. Many of those Compet-n players who used regular Doom2.exe for Final Doom demos implied they were using the Final Doom exe. No idea why. Maybe they didn't understand the difference if they didn't have a real copy of Final Doom, or they just copied the txt template without ever modifying it.

 

[ZeroMaster010]

I swear there's an infinite number of desyncs in the videos I uploaded to the youtube channel for TNT and plutonia. Those videos were made with a batch file set to run through a set of demos (all uv-speed for TNT etc) and complevel was probably set to 4 for all of them, not sure. And as grazza mentioned they will play back fine with complevel 2, a lot of the videos were corrected afterwards, but there always seem to be some I missed.. so I'll try to upload/correct the ones that desync some day.

 

... oh and as you can see at the end of the video I actually noticed and exited the demo when making the video, just completely forgot about it :)

[Bob9001]

TNTR

 

My Episode 1 Nomo and Nomo 100% are listed as a D2all when they should be listed as Episode 1 :)

[GarrettChan]

On 5/31/2020 at 9:19 PM, Grazza said:

This is a -complevel 2 demo and it plays back fine.

Please correct me if this is wrong. According to the info I read (it's very difficult to find things to read about -complevel TBH), -cl4 and -cl2 are mostly the same, but -cl4 has a glitchy teleporter which throws the player all the way up to the ceiling. I realized this when I did runs on -cl4, but I know nothing about anything else. I guess maybe the demos desynced because the extra time the player will take by falling from the ceiling to the ground.

 

I actually wanted to know more about these for a long time.

[Grazza]

Two differences: the teleporter stuff you mention (which can cause desyncs, but not inevitably - presumably it can affect what things see the player and when; if nothing sees the player falling, then perhaps all is well, I suppose, but maybe someone has researched this properly and knows more) and -complevel 4 has the same lost soul behaviour as -complevel 3 (which can more easily cause desyncs).

 

BTW, if anyone doesn't know how to enable autoloading, start up prb+ with the built-in launcher (if you've disabled it, then shift-dblclick on the exe), and in the "commands" windows, select the ''...with DOOM demos (-auto mode)" option. (This should really have been made the default behaviour.) Then most demos will play back correctly just by clicking on them, either because (for older demos) the file-name is recognized or because it has the demo footer (for newer ones where the player has this option enabled). You can also disable the launcher or set it to smart, as otherwise it just gets in the way when playing back demos.

 

ZM: if you want to get a list of which Compet-n Final Doom demos need -complevel 2 to avoid desyncs, then the autoloading patterns are a handy reference. They are in glboom-plus.cfg. Patterns 26 ("DOOM 2: Plutonia pwad"), 28 ("DOOM 2: TNT - Evilution pwad") and 29 ("DOOM 2: TNT - Evilution pwad with map31 patch") are the relevant ones.

[GarrettChan]

33 minutes ago, Grazza said:

if nothing sees the player falling, then perhaps all is well, I suppose

I realized the player will be locked in place for a little bit when you teleport, so this falling doesn't actually lose you time, perhaps?

[vdgg]

@GarrettChan: The demo by Archy, pt11-420 is 4:21 if you use -complevel 2 by mistake :-P

 

And, as said, lost soul behaviour is a much more likely and important case for desyncs in general.

[Xeriphas1994]

10 hours ago, GarrettChan said:

I realized the player will be locked in place for a little bit when you teleport, so this falling doesn't actually lose you time, perhaps?

 

Unless the falling time is longer than the lock time would have been, resulting in the player thing not responding to the first few recorded movements after teleportation, and then it desyncs. pl06-322 is an example noted during PrBoom development.

 

ISTR even landing below the floor costs one tic as the player height is "corrected" (like wallrunning), but I don't have code to support that.

[Coincident]

All of the demos submitted in April were uploaded to DSDA.

 

Please verify that all of your demos are in order. Post here or PM me if there's anything missing, or any mistakes.

 

If everything is OK, we'll continue to upload demos for the month of May soon.

[DuckReconMajor]

Yay! Thank you.

 

Both mine and j4rio's demo here should have the 

"Plays back with cc2_m24f.wad."

notation.

[Coincident]

On 6/8/2020 at 5:22 PM, DuckReconMajor said:

Yay! Thank you.

 

Both mine and j4rio's demo here should have the 

"Plays back with cc2_m24f.wad."

notation.

Done. We will proceed with the demos of May shortly.

[Never_Again]

On the Vilcore page all of MT's demos as well as Opulent's MAP32 one are for the original version (2000) and desync with the final version (2003) since the latter had the nodes rebuilt on all maps. It would probably be best to split off the 2000 version with its 34 demos.

MT's 15:00 MAP32 run needs the "Plays back with map32.wad" comment.

[Xymph]

In the players list, a dummy entry https://dsdarchive.com/players/tatsurd-cacocaco popped up adjacent to https://dsdarchive.com/players/tatsurd_cacocaco .

[Xeriphas1994]

https://dsdarchive.com/files/demos/doom/47139/e1m1-010_Loomeh.zip

 

This is skill 3. Count the zombies in the computer room.

[tchkb]

https://dsdarchive.com/wads/stardate

My D2ALL HNTR-Max run is listed simply as "Skill 2" instead of "Skill 2 Max". I think I've already pointed it out at some point, but maybe I'm just imagining things.

[Xeriphas1994]

F1M1D055.lmp has length 56.91. (I wrote a script. Duck and cover.)

 

Also, this implies milliseconds will appear unconditionally (at least within a given category). Is that the plan? I don't have a strong opinion like the people upthread, but the current situation does put certain runs out of order, e.g. kraflab's MAP01 Reality is 54.97 but sorted as 54.00.

 

Thanks again for being smarter than me and organizing so many amazing performances. I hope these posts never sound like cheap shots! It's just that the site looks SO GOOD and I want it to be even better.

[GoneAway]

Lots of runs are missing tics - this will be fixed eventually, though as you might expect we prioritized getting tics for runs that are records and have ties.

[siteod]

so uh, does editing category filters really still not do anything, or is my browser and/or my security extensions just not working with the site correctly? no shade, ive just been away from the demo scene for a while and not sure whats up

[Xymph]

The latest entry for Jakub Puk on AV MAP12 states to have used the DooM engine, which should probably be DooM2.

[Dimon12321]

Wow, it's a new one! Now multimap replays have individual timings for each completed map.

Spoiler

 

[Bob9001]

25 minutes ago, Dimon12321 said:

 

  Hide contents

 

I noticed this awhile ago, but didn't say anything since I thought it was there from the start.

 

Nice find!

[GoneAway]

On 6/22/2020 at 2:16 AM, siteod said:

so uh, does editing category filters really still not do anything, or is my browser and/or my security extensions just not working with the site correctly? no shade, ive just been away from the demo scene for a while and not sure whats up

That's correct. I haven't looked into it yet since it's quite low priority for me. But if this is something there is more immediate desire for then I can shuffle priorities.

 

Dimon / Bob: this was always there (although it is probably not available for every run).

[Xeriphas1994]

Is the "dubious" icon appropriate for plagiarism?

 

   Hold on - "William Wheeler" - You're the infamous .LMP THIEF !

   You stole (NM!) .LMP's from Steffen and Anthe and claimed them
   as your own... Congrats: you've just gained # Penalty Points !

   So who knows who the true author of this .LMP is anyway ?

LMP Title               : DOOM ][ ... The Movie
Filename                : doom2-nm.lmp

[ Just happens to be byte-for-byte identical to the one Steffen made ]

Author                  : William Wheeler

Description             : I beat the game on NightMare. About time.

[ Liar ! ]