The DWIronman League dies to: Vae Victus 1 & 2

[Dime]

www.twitch.tv/kingdime

 

Vae Victus Ironman, Will Malcom and Dumont finish off canada or can i sneak ahead of alfonzo's survival time.

[fabian]

On 1.5.2017 at 8:34 PM, NecrumWarrior said:

So, the complevel being Doom(Strict) this month, is a Vanilla style port like Crispy Doom an acceptable way to record my demo for the attempt, or should I stick with PrBoom+?

By design it is impossible to configure Crispy Doom in a way that results in it being vanilla incompatible when recording or playing back demos.

[Alfonzo]

Updating the OP shortly.

 

Quick question. Both Dime and ella guro have run into issues with loading the vv1 wad files (the base and the provided fix) correctly, and I'm wondering if it's fair to penalize players who make the error of running them in the wrong order, even provided instructions at the top of the page. Would it be better to continue to use a 'patch' file like this one, or should fixes be integrated into a custom 'version' of the base wad, requiring that players load just one file instead of two? My reasoning at the start of the month was that players might already possess the base wad or want to keep it for normal use after download, and not want to have an additional file of comparable size on their HD that would only be needed for running old ironman demos. 

 

Anyway, I'm disappointed that Dime didn't have a proper crack at vv1, because he cleaned up in the sequel and was on track for a good finish. 38:58! Solid stuff.

 

On 5/4/2017 at 7:07 AM, RjY said:

Please note: to play the VV+VV1fix demo, PrBoom+ must have reject overflow emulation turned off, otherwise it will desync on map03. Apologies for the inconvenience. I'll write a technical explanation, if required.

It may be required! If it's due to an error I made in putting the fix.wad together (i.e. having reject emulation turned on wouldn't ordinary run into conflict with any appended file or something) then second tries are in order.

Edited May 7, 2017 by Alfonzo : I am not a double post.

[Havoc Crow]

Both demos together: http://www.mediafire.com/file/r4ok5drq4to3nvq/DWIronman10_JudgeDeadd.zip

 

VV1: Died on MAP02 after I stood next to a barrel I hadn't even noticed.

 

VV2: Died on MAP07 in the lamest way possible, falling into lava with no way to leave :< SO CLOSE!

[BigDickBzzrak]

6 hours ago, Alfonzo said:

vv1-fix stuff

It would've been easier if you had a command-line example in the OP that shows in what order should the files be loaded in and all that. I imagine it would make for far less runs disqualified due to "illegal comp settings". Just my 2 cents of course.

[Rathori]

IMO patch WADs are better, exactly for the reason you've stated: having a separate version of WADs just for Ironman would create a mess of WADs, and, personally, since pretty much every Ironman WAD is new to me, I put the files I download here into my WAD collection folder for a full playthrough later on and for watching DSDA demos after I've beaten it.

I think the issues with people accidentally loading the WADs incorrectly can be easily avoided by providing full command line required to start the WAD properly in the OP like this:

prboom-plus -file vv vv1-fix -complevel 2 -skill 4 -record my_demo

This should also take care of wrong complevels and difficulty settings being used.

 

BTW, I couldn't find the correct loading order anywhere in the OP. It only says that vv1-fix.wad corrects some issues, but not that it should be loaded after vv.wad

[Alfonzo]

Thanks for that, you two. Makes sense. I'll code it into the OP and then place the onus on the players, I think.

[Dime]

Even if i spectacularly survived Vae Victis 1, I doubt it would be much faster than a 1:35:00 or 1:40:00 + compatibility issues on map 3 regardless. Having any previous experience is most helpful on map 6 VV1. I started to explore a bit after dying and still didn't really figure out the progression for caverna fiesta.

[RjY]

5 hours ago, Alfonzo said:

It may be required! If it's due to an error I made in putting the fix.wad together (i.e. having reject emulation turned on wouldn't ordinary run into conflict with any appended file or something) then second tries are in order.

Summary of summary: ZDBSP default output is unfortunate for vanilla compatibility, and I (and my source port) disagree with PrBoom+ on how to handle the consequences.

 

Summary

• At some point ZDBSP started writing, by default, zero-length reject lumps. (I infer you used ZDBSP with default settings to build VV1fix.)
• This is incompatible with vanilla.exe treatment of "short" rejects (i.e. not enough bytes to store square-of-number-of-sectors bits), which reads rubbish from memory past the end of the lump.
• My engine in vanilla compatibility mode differs from PrBoom+ in handling zero-length reject lumps; I don't think what PrBoom+ does with missing rejects makes much sense, as it doesn't fix actual vanilla.exe compatibility (this is impossible when a reject is more than 16 bytes short).
• For the widest portability of a "Doom+" wad, a zero-padded (not zero-length) reject table should be written (if not a properly built reject).
• Alternatively you may decree conformity to the defacto standard is required; or just that I stop recording with a custom engine.

Background

 

Reject lump calculation can take considerable time. Thus, traditionally most nodebuilders would write out a reject lump of the correct size but filled with zeroes. This should have no effect on gameplay other than framerate, and a true reject build can be delayed until the wad is ready for release. Or in many cases, especially on modern hardware, the reject build step can be omitted completely as it does not result in a noticable performance improvement (monster line-of-sight checks comprise little of the workload unless you have thousands of monsters in a huge map; a notable exception was okuplok_untitled's 20K+ monsters placed in disjoint arenas gaining a massive improvement after being run through zennode-reject).

 

If a proper reject is found to be ineffective, the zero-padded reject lump can be a waste of resources. Using disk space especially with large maps (3000 sectors => 1MB block of zeroes), memory used by the engine, time spent reading from said memory (cache misses), etc. So a shortcut taken by ZDBSP is not to write any data, and leave the lump empty (zero-length). Obviously this requires the source port to detect this and handle it. From ZDoom point-of-view this is a no-brainer. Update nodebuilder and engine, release both simultaneously; everyone upgrades to latest, carry on. Indeed afterwards most current engines were patched to be aware of the zero-length trick. It is a simple change to make, only a few lines of code.

 

But it's bad for portability to other engines which aren't updated similarly (and may never be -- of course the case with vanilla). Non-aware engines will read rubbish from memory past the end of the lump, leading to it being essentially random as to whether two sectors can see each other. If the lump is only short by up to 16 bytes, entryway discovered the random rubbish is predictable, and wrote code to emulate it. But trying to emulate a reject more than 16 bytes short is not possible in the general case. However, this code as written, still activates, even though the resulting recording will most likely fail to play in vanilla.exe, or conversely a vanilla recording on a map with too short a reject will fail to play in a source port.

 

In contrast, in my view it is obvious that a zero-length reject lump means an intentionally vanilla-incompatible map, meant for at least limit-removing source ports which adopt the ZDoom behaviour (a zero length reject lump is shorthand for a sufficiently large block of zero bits). Thus, rboom (my PrBoom v2.3 based engine, with a user interface tweaked to my preferences to the point that I hate using anything else) does not try to emulate the overflow in the particular case of an entirely missing reject lump. However, PrBoom+ by default in -complevel 2/3/4 (vanilla behaviour emulation but rendering limits removed, aka "Doom+") will still emulate the overflow as far as it is able. This, as I explained, is, in my opinion, pretty silly in the case where the lump is missing/empty (what to do with existing but truncated data too short to emulate is less clear and more debatable).

 

This could have been avoided if VV1fix had been built with a zero-padded reject table, but that's not the default behaviour in ZDBSP, you have to add a command line option for it (naturally ZDBSP's defaults are oriented towards ZDoom, not vanilla). In fact PrBoom+ will play my demo if you run "zennode -n- -b- -rz vv1fix.wad", to insert zero-padded rejects without touching the rest of the map lumps. I think if you had built a zero-padded reject in the first place, that would have been better for maximum portability. Yet, on the other hand you may argue that what PrBoom+ does is the de-facto standard, and I should suck it up and make rboom fall in line -- or even that I shouldn't be using a non-standard port to record demos. Especially for a competition. Although to that I would say, I am unlikely ever to progress higher than mid-table mediocrity, let alone win. So please at least let me have fun. Indeed, I used to write long disclaimers in my DSDA text files to the same effect, but nobody ever really objected, so I stopped.

 

[WH-Wilou84]

VV : Died in map 07 due to infinitely tall corpse shenanigans (and eating a rocket). :/
I used both vv.wad and vv1-fix.wad, in this order.

VV2 : Finished in less than an hour.

Good picks as always, Alfonzo. :)
Vae Victus is a nice "Deus Vult Lite" kind of mapset, and my favorite of the two. Both mapsets would have benefited from more playtesting, though.

VV_VV2_Ironman_wh.zip

[skillsaw]

https://www.twitch.tv/videos/143749292

 

VV2 is a pretty fun mapset. I don't have much to say about VV1 because, uh, reasons :)

[Ancalagon]

I missed the part where VV is like deus vult.

[Suitepee]

https://www.twitch.tv/videos/144619814 = 13 maps cleared; survived VV2, died on map 7 of VV1.

 

Infinite height hanging corpses on map 7 killed me. (well technically the Cyberdemon did, but still...) Much cursing followed suit. I did however manage to survive map 6 reasonably well, which was by far the toughest obstacle of this run.

 

Vae Victus 2 barely posed a challenge to be honest, but the levels looked nice.

 

With this result I now regain my lead and go 3-2 up on an_mutt in the 2017 Britbowl.....I look forward to next month's instalment. (especially since I'm now on a Doom wad break for a couple of months, meaning that the Ironman wads are all I will play for a bit)

[TMD]

Won't have internet access this month...as far as I know.  I followed advice and got the heck out if my current situation, but it leaves me without wifi until June

 

 So consider this entry an experiment in demo posting.

 

See you next week :)

[Alfonzo]

16 hours ago, Suitepee said:

13 maps cleared; survived VV2, died on map 7 of VV1.

You did all the hard work to beat map06 and deserved victory over VV1, for sure (same to you, Wilou). It's a real shame it ended the way it did, and I feel bad! Take consolation in knowing that this is still your best performance this year, even if you didn't really 'survive' the challenge.

 

On 5/7/2017 at 2:04 AM, Dime said:

www.twitch.tv/kingdime

Quit being all lazy and highlight your runs so I can link to them properly in the OP without them expiring! I can't make a habit of hunting down your deaths like a Doomed Debt Collector, anyway. That sentence is fun to say.

[SilverIce]

Vae Victus: A Hell Knight and I killed each other on MAP02.

Vae Victus 2: An unfortunate incident with the rocket launcher softened me up for a Revenant homing attack on MAP05.

 

 

 

This must be my punishment for having the gall to survive last month's session. :)

 

[Roofi]

VV2 done in : 56:24 but....

 

died at VV1's map 05 because of dat fu**** cyber which didn't stop spamming rockets to me. :D

 

 

 

 

Roofi_Ironman_VV.zip

 

I love both wads , especially hellish maps because the architecture reminds me a lot of Alien Vendetta.

Edited May 20, 2017 by Roofi

[elderdragonbrasil]

OMG! How could someone done it on blind under 2 hours?!

[Demon of the Well]

Here's me again: With apologies to Cynical...

 

Finished the original Vae Victus in 43:41, and the sequel in 30:34, played in separate sessions over the past couple of days. Almost died once in VV2 eating a point-blank rev-rocket to the face in m04, but other than that it's relatively smooth sailing, both because of the positive glut of ammo and items that continuous play eventually affords in both sets, and because so much of the combat in the later maps is so easily skippable, making these WADs highly amenable to all sorts of darling, snuggly-wuggly Baby's First Speedrun Hallmark moments (as my own runs attest). Progression is the real bugbear in this case; the cavern switch in VV06 is the main offender, but VV2 in general is full of somewhat convoluted door/switch/key combos which eat up time bit by bit.

 

For context: I had said earlier in the thread that I had played VV before, but not VV2. This was simply incorrect. When I went to play VV2, I recognized it as soon as I heard "Beginning" from Castlevania III start up in m01, and remembered most of the WAD fairly well thereafter, having most recently played it as a 'random' selection from my WAD collection just a few years ago. I apologize for the misleading comment, it was not my intent to be sneaky. Seems I had VV2 mixed up in my head with "Aeternum" by Skillsaw (similar AV references, orange sky, general texture schemes, etc.), and probably some other more or less similar WADs from around the same time as well; ultimately the original VV.wad is a significantly more memorable set for one reason or another (even though it's also significantly less polished), and I had thought at the time that VV2 would be more like it than it in reality actually is, I guess.

 

This was a weird one. I think the particular designs in these sets actually make it more likely you'll die if you play carefully and responsibly in the endgame maps than if you refuse to show the maps any real respect (though foreknowledge is a big swing force as always, of course), which is perhaps somewhat counterintuitive. Though some hardcore types might well argue that this is actually the case deceptively often with more tightly designed sets, granted, the rub here is that VV's really not particularly tightly designed in that regard, though of course community standards of normalcy have positively skyrocketed lo these past few years, I suppose. Nevertheless, while not the most elegant things in the world, some of the specific spot-references to Deus Vult and Alien Vendetta in these sets are heartwarming, particularly the affectionate Johnsen-isms and Malde-alikes at the tail end of the latter.

Edited May 27, 2017 by Demon of the Well

[Cynical]

Fake news.

[Demon of the Well]

* Fair and Balanced

[Suitepee]

So I wonder what's on the agenda for next month's Ironman? Did DOTW beat Veinen's time?

[Scotty]

Typically i started to feel very comfortable - and that's always when it all goes to shit. VV2 tomorrow.

 

https://www.twitch.tv/videos/148313573

 

edit - VV2 @Alfonzo:

 

god fucking damn it

 

https://www.twitch.tv/videos/148534027

Edited May 31, 2017 by Scotty

[Rathori]

A bit of a late entry, but I promise it was still May when I started recording the last demo.

Dead on map 05 in both WADs: On May the 1st in VV1 and on May the 31st in VV2, both are very stupid and frustrating deaths in easy parts where I had no problem in my original playthroughs.

rth-iron-vv-1.zip

rth-iron-vv-2.zip

[Pirx]

On 5/26/2017 at 5:13 PM, Demon of the Well said:

Here's me again: With apologies to Cynical...

[snip]

 

[ClonedPickle]

it's still may somewhere

 

e: https://www.twitch.tv/videos/148664715

 

double map05 deaths

Edited June 1, 2017 by ClonedPickle