For a forum that prides itself on trying new things, for the love of all that is novel and interesting, please can we have a new QotM? - Krill

Create an account  

 
Reversing to Orion - project 1oom

I got the "trade bug" where I was able to offer techs that I don't have again, but I wasn't able to reproduce it in a save file, probably because of the non-deterministic way tech trades are selected.
Reply

If you try to invade with, say, 50 transports, and 10 get through, the message will read "10 out of 10 landed" instead of "10 out of 50"
Reply

Graphical glitch when voting for meklar

https://www.dropbox.com/s/oa3gzbrmc1eh6b...1.bin?dl=0

Will sometimes crash with non-deterministic. I'm enabling all optional fixes except the waste_calc_fix. Can also try hailing Klackons to reproduce tech trade bug with relatively high success.

Edit: And I can attack a Meklar colony after which the Klackons declare war for "unprovoked acts of war"
Reply

The project coder replies:

(September 19th, 2018, 23:04)Jeff Graw Wrote: error: patch file 'fixbugs.pbx' item 21 numid 'waste_calc_fix' (0) invalid!

Windows 10, 0.8-21

Can't reproduce. The attached fixbugs.pbx works for me. 1oom_log.txt and pbx or it didn't happen. This looks a lot like "used v0.8 by accident" to me, but it wouldn't be the first weird memory corruption issue. I'll grab the win32 build and mess with wine while waiting.

Please paste the full version string from 1oom_log.txt when reporting bugs. The missing part is there for a reason. It allows me to go instantly to the relevant commit. I omit it in these posts because for those who are not compiling 1oom from git it's just noise and due to my workflow it's likely that I mess it up. Also, copying the version from the same log as the error gives a chance to notice unexpected versions.

Quote:I got the "trade bug" where I was able to offer techs that I don't have again, but I wasn't able to reproduce it in a save file, probably because of the non-deterministic way tech trades are selected.

The tech trade selection is deterministic (assuming determinism is not disabled with the PBX num "determinism"). If you have a save and do everything the same way/order, the list remains constant. The (optional) non-determinism comes from sitting idle.

Quote:If you try to invade with, say, 50 transports, and 10 get through, the message will read "10 out of 10 landed" instead of "10 out of 50"

Fixed in v0.8-23. It took longer to find a suitable save to test it than to actually pinpoint and fix the bug. I bet it took more time for me to write this paragraph than it would have for you to attach/upload the save.

Quote:Graphical glitch when voting for meklar

Can't reproduce and/or see it. Please grab a screenshot (Ctrl-F5) and elaborate.

Quote:Will sometimes crash with non-deterministic. I'm enabling all optional fixes except the waste_calc_fix. Can also try hailing Klackons to reproduce tech trade bug with relatively high success.

If you get a crash, backup the 1oom_save8.bin. (Have I written that somewhere before?) If you're disabling determinism for some reason (which I'd love to hear), it should be reproducable with "1oom_classic_sdl1 -file foo.pbx -load crash_backup.bin -nextturn -oldsaverng" if battles are Auto'd away at roughly the same speed.

Quote:Edit: And I can attack a Meklar colony after which the Klackons declare war for "unprovoked acts of war"

Save (and pbx) please.

Thanks for the reports and for making a compelling case for removing the determinism toggle.


Attached Files
.zip   fixbugs.zip (Size: 679 bytes / Downloads: 0)
Support Free Software projects!
Reply

"Unprovoked acts of war" was the turn right after the one in the save game, sorry, forgot to mention.

Seems like you were right about using an earlier version of the game. Looks like recent Windows development builds no longer contain an SDL2 executable, and I failed to notice that when overwriting the previous files in the 1oom directory.

Tech trading didn't feel deterministic, but maybe I need to look at it again with a recent executable.

Regarding determinism toggle, I used that to avoid a nova event that took ~2X (rough estimate) the maximum amount of RP I could generate at a planet in the same time frame after changing fully to research (minus going into waste) the same turn, and maxing out stimulus there every year at that. Pretty sure the colony was at population capacity too, but was poor and didn't quite have full factories. Now that I think about it, this may well be a bug. But if it is, only one that I encountered on an old executable.

I wouldn't remove the determinism toggle though. Determinism by itself defacto violates your philosophy to be true to vanilla, but is a necessary evil. Like the redirect "bug", save scumming is a thing that a lot of players do, and they're used to a certain behaviour. That behaviour can, and for that slice of the playerbase certainly does, contribute tangibly to the gestalt of the game. There's absolutely a compelling case, at least in the project's present state, for having determinism on by default to help with development, but it's one thing to disable vanilla behaviour, or at least vanilla behaviour that can really be felt by players, and another to remove it entirely.

As an aside, determinism doesn't necessarily solve save scumming, but it does transform it. Instead of save scumming for a breakthrough or a correct response in diplomacy, a player can save scum to pick the right field after an espionage roll or reduce research spending during a turn they aren't going to make a breakthrough anyway, etc. Eg. relying on deterministic behaviour rather than exploiting uncertainty.
Reply

The grumpy project coder replies:

(September 21st, 2018, 12:04)Jeff Graw Wrote: "Unprovoked acts of war" was the turn right after the one in the save game, sorry, forgot to mention.

I tried to attack a planet and got the Alliance with Klackons broken, which is not the same thing. During that turn you may have done any number of things that could have affected the outcome.

I still need that save to fix it. Reporting a bug without the exact save to reproduce it is a valid alternative to pissing on my morning coffee.

Quote:Seems like you were right about using an earlier version of the game. Looks like recent Windows development builds no longer contain an SDL2 executable, and I failed to notice that when overwriting the previous files in the 1oom directory.

I was not aware of this. I'll ask what's going on. EDIT: The issue has been fixed and the SDL2 exes are back.

Is the graphical glitch gone now?

Quote:Regarding determinism toggle, I used that to avoid a nova event that took ~2X (rough estimate) the maximum amount of RP I could generate at a planet in the same time frame after changing fully to research (minus going into waste) the same turn, and maxing out stimulus there every year at that. Pretty sure the colony was at population capacity too, but was poor and didn't quite have full factories. Now that I think about it, this may well be a bug. But if it is, only one that I encountered on an old executable.

Thanks for clarifying. I'll take a look at the nova event code.

Quote:I wouldn't remove the determinism toggle though.

Good points, I'll keep it.

The main reasons for determinism are bug reproducability and multiplayer streamlining. For example, ground battle RNG was influenced by when the player clicked to skip to the end of the battle. Removing this nondeterminism allowed to run all the battles at once and send the starting conditions (including RNG state) to the all the players to view at their leisure (or in the case of hotseat: without excessive seat-switching).

I do not care about save scumming. What people do with 1oom for entertainment is none of my business.
Support Free Software projects!
Reply

(September 21st, 2018, 15:50)1oom_aaron Wrote: I still need that save to fix it. Reporting a bug without the exact save to reproduce it is a valid alternative to pissing on my morning coffee.

Sorry again, what I meant was "just hit turn twice." For reference, if I send a save that contains multiple bugs on concurrent turns, and I don't mention anything else, it means you just need to hit turn to reproduce.

Using an actually up to date binary I no longer get the "unprovoked acts of war" message, although they do declare war. Probably a coincidence.

I no longer get the same graphical corruption when receiving a diplomacy hail either.
Reply

The grumpier project coder replies:

(September 21st, 2018, 22:20)Jeff Graw Wrote: Sorry again, what I meant was "just hit turn twice." For reference, if I send a save that contains multiple bugs on concurrent turns, and I don't mention anything else, it means you just need to hit turn to reproduce.

Nearing your Dropbox quota or something? Please send individual saves in these cases too. I can't reproduce the bug as you haven't stated the old version (and PBX) you were using.

Quote:Using an actually up to date binary I no longer get the "unprovoked acts of war" message, although they do declare war. Probably a coincidence.

State the old version (and PBX) you were using and maybe the message is reproducable. I didn't get a war declaration. If you think that is a bug, please send the save. This coffee sure tastes funny.

Let's clarify. When you notice a bug that happens during turn processing, make a backup of 1oom_save8.bin immediately. Try to get it to trigger again. If it does, report the bug with the following info:
  • the version from 1oom_log.txt
  • description of what went wrong
  • PBX file (if any); "using fixbugs from the same version (with this small edit)" is fine
  • the backup save
For non-turn processing bugs (such as audience weirdness), include a non-undo save from the turn and the steps to reproduce. If at all possible, please report bugs at gitlab since it is far more convenient for me.

Quote:Graphical glitch when voting for meklar

Quote:I no longer get the same graphical corruption when receiving a diplomacy hail either.

Diplomacy hail. Guess who scrutinized the council code and tried to spot glitches when "voting for meklar"?

Looked at the nova code. Found no bugs. Updated OSG Errata.
Support Free Software projects!
Reply

I finally got around to trying this out. Better late than never, I guess! I only played a little bit so far, though.

- It was easy to build on OpenBSD! Great work making it portable! I'll try to get it added to the package system when it's done.

- I can't get the music to sound the way I like it, and the -sdlmixersf option didn't help. With MOO1 in DosBox, I use a virtual MIDI interface to pipe the MIDI from DosBox into fluidsynth. Is there some way to add native MIDI output as an option with the SDL libraries?

- I like to see the Planetology research options after my first turn, so I set the Planetology slider to 100% and invested 1 bc from my homeworld into tech. This works in MOO1, but in 1oom I have to invest 3 bc in a single turn before I can choose a tech.

- Finally, what's the preferred way to ask questions and give feedback? Here in the thread, or would it be more convenient on GitLab?
Reply

(September 22nd, 2018, 22:07)1oom_aaron Wrote: I can't reproduce the bug as you haven't stated the old version (and PBX) you were using.


State the old version (and PBX) you were using and maybe the message is reproducable. I didn't get a war declaration. If you think that is a bug, please send the save. This coffee sure tastes funny.

I don't think it's a bug. In fact, I've seen the same message pop up in other scenarios and it just appears to be one of the random strings that gets pulled when an empire declares war for any reason. Just an unfortunate intersection of my memory of the messages being bad, and the message wording looking like it could apply to some other event.
 

(September 22nd, 2018, 22:07)1oom_aaron Wrote: Diplomacy hail. Guess who scrutinized the council code and tried to spot glitches when "voting for meklar"?

To be fair, an explicitly verbose description of a bug, while nice, isn't totally necessary when the reproduction steps are infallible and the issue is obvious to behold. It's unfortunate that the stealth removal of the SDL2 executable was a thing here. From my perspective at the time, it was largely reasonable to assume that you'd see the same thing on your end as I on mine.


(September 22nd, 2018, 22:07)1oom_aaron Wrote: Looked at the nova code. Found no bugs. Updated OSG Errata.

Odd that I can't remember running into the same situation in vanilla. It's entirely possible that my memory is bad, or that I failed to get so unlucky previously. Will it be getting a general and/or PBX fix anyway? A nova event that is literally impossible to fix before time runs out strikes me as something that would be universally unwanted by practically everyone, regardless of whether it exists in vanilla or not.

This talk gives me another idea as well. Perhaps the amount of RP required / turns left could be displayed somewhere on the colony panel (eg. close to "Nova") when uiextra is enabled?
Reply



Forum Jump: