Matrix Games Forums

2nd turn after switching from 1.07.01 rel. to 1.07.03 Matrix.

Entertainment Subscore drops from 17 (50/3=16.67) by 5 points to
12 (not 50/4=12.5 what should be rounded up to 13?!).

Health Subscore stays the same as before (50/3=16.67).

No cause visible - Casino (private) works fine and produces 50 qol_entertainment.
These are shown on the Civilisation Score table.

Only peculiarity I can find:
Arena (private) started building this turn.
Obviously does not yet produce qol_entertainment.

BlueGreen autosave_round38.se1
BlueGreen039.se1

... waiting for zgrssd to come in and tell me that this is not a bug and that I should read the manual and that I surely upgraded my town to 4 and deactivated the casino and whatever ...

10
.
9
.
8
.
7
...

It is in the turn 39 savgame.
But doing a turn resolution from turn 38 endturn in 1.07.03 does not replicate it.
I happened to have a savegame of the same scenario version (1.07.01 #39), also no replication on End-Turn.

That would mean it is either a race condition, a installation issue or something else specific to how the turn 39 savegame came to be. Fascinating.

Does the bug replicate for you when pressing end turn (from your own savegame) ?

best wishes,
Vic

I loaded and ran "BlueGreen autosave_round38.se1" several times.

On the 3rd try, it happened again - somewhat:

BlueGreen039SaveAfterRerun3.se1

I noticed on these reruns that no time the autosave processes identically:

[*] Game does a new autosave before processing. This autosave is not of the same size as the old one (Might be a version issue?)
[*] turn might be processing slightly faster (1:55 vs. 2:01 min.) when re-loading after "quite game", instead of loading after completly re-starting the game. Would need much more data (statistics), don't have the time for that.
[*] private econ build target seems to shift often, before the asset is acutally build (e.g. turn X - build target Y, next turn Z is shown build); I'll try to test + document this.[/color]
[*] ... more details to come if I find the time to properly test and document ...

edit

I noticed on these reruns that no time the autosave processes identically:

Game does a new autosave before processing. This autosave is not of the same size as the old one (Might be a version issue?)
turn might be processing slightly faster (1:55 vs. 2:01 min.) when re-loading after "quite game", instead of loading after completly re-starting the game. Would need much more data (statistics), don't have the time for that.

Cross check from my side:
Downloaded Turn 38 autosave: 2922263 Bytes
Turn 38 autosave from previous test: 2922291 Bytes

Downloaded savegame again, then re-used previous attempts savegame:
1st Try: 2922291 Bytes, no noticeable QOL oddity
2nd try: 2922291 Bytes, no noticeable QOL oddity
3rd try: 2922291 Bytes, no noticeable QOL oddity
4th try: 2922291 Bytes, overall QOL saw a increase to 8

That the game is creating a new savegame is a known behavior. I reported it as a bug actually.
I could find no size aberations in the new savegames in my tests.
AI movement behavior seems deterministic
Stratagems, finishing of the Model and Events/Reports are non-deterministic
I noticed only the one increased overall QOL value.*


*Some systems roll (value)dice to figure out the effect and/or change. Spies are most noticeable, but I think the Unrest and Danger impact on happiness does as well. No idea if such a effect for QOL change is intentional.

ORIGINAL: zgrssd

...
4th try: 2922291 Bytes, overall QOL saw a increase to 8

Are you speaking about the "QOL Score" or the "Civilisation Score"?
Did the game show the CS go from 5 to 8 without changes in QOL-Score, or did the QOL-Score decrease to 8 from 18?!?

ORIGINAL: Arralen

ORIGINAL: zgrssd

...
4th try: 2922291 Bytes, overall QOL saw a increase to 8

Are you speaking about the "QOL Score" or the "Civilisation Score"?
Did the game show the CS go from 5 to 8 without changes in QOL-Score, or did the QOL-Score decrease to 8 from 18?!?

Right, my bad. Civilisation score was usually static on 5 both before and after the end turn.
But once it jumped to 8 after end turn. I have yet failed to repeat it.
And after about 3 more tests, I had the Health Score drop from 17 to 14.

It smells like a race condition in that code somewhere. I guess this change: "-Made some relatively good speed improvements to turn speed. However this also made me change some compilation settings and might give some unwanted bugs, didn’t find any, but lets see. Stay aware this is an open BETA. capitals for emphasis. " opened some code for race conditions

I'll be switching the next open beta .04 back to different compilation mode. Maybe you guys would be so kind to test if this problem stays with .04 or if it goes away?

Best wishes,
Vic

Bump.
Anybody to be willing some tests on those savefiles would be much appreciated!

It is important some other systems run some tests on this as well.

As the cause it either structural OR caused by new compilation method in .02, but switched of again with .04

So if anybody can still repeat this bug report with .04 or .05 that means that the issue is not coming from the speed increases made with .02/.03

I had some issues testing it, as I also had the CTD bug. Nothing on the first turn after the .05 patch, but I will try to keep an eye out for any aberations.

Looking into this, what seems to be happening is that there are two things coming together:
1. Some kind of "intrinsic" increase/decrease depending on the points.
2. This is modified to some extent by governor rolls.

What I don't quite understand is how these two parts interact.
Increase/decrease:

The rolls:


Different try:


Note the horrific investigation roll, but the security score went up more than the previous one.
Also, the science roll was quite bad, but the education score went up by the largest amount of all tries.

Different try again:



And a last try:


A terrific science roll, but the education score went up by the smallest amount!

It feels like the governor rolls change the limits of the scores allowed? Which is why the entertainment score drops so much when the governor makes bad rolls. But the rolls do not seem to have an effect (or maybe an anti-proportional effect?!) on the actual increases.

Thanks for all the feedback Arralen, Zgr and Turtle!

I figured it out. There is a very small absolute boost for the health,entert,edu and sec scores based on skill rolls for governor (so if there are now assets some points can still be generated). But it was malconfigured and allowing negative changes.

ORIGINAL: Vic

Thanks for all the feedback Arralen, Zgr and Turtle!

I figured it out. There is a very small absolute boost for the health,entert,edu and sec scores based on skill rolls for governor (so if there are now assets some points can still be generated). But it was malconfigured and allowing negative changes.

I was worried that might not have been "it" after all, as I had no repetition in .05. Until this turn.
I had:
Health Score: 58, (-5). Health Points: 250 (+/-0)
Security Score: 53, (+2); Security Points 292 (-4)

This should be fixed with .06

.05 was an emergency patch to fix the crashes.

Heh, I've noticed it too, but then considered it as annoying, but "working as designed"...

ORIGINAL: zgrssd
AI movement behavior seems deterministic

No it isn't (at least not for attacks), I've seen several times the AI deciding to finally not to attack / attack the 2nd time I processed the turn, not the same thing it decided the first time.

By the way, could we please have a "fixed seed" option ? Would help a lot for testing & learning...