v2.1.1 Bugs Thread

[Steve Walmsley (OP)]

Assuming 1000 stars, with Local Chance at 50 and Local Spread at 15, there is a 50% chance that the JP will connect to a system number within 15 of the start system. That doesn't mean it will connect to an existing system - just to that number. It will only connect to a system if one is already occupying that system number slot.

There are also some restrictions on connections. A jump point can't connect to a system if the two systems are already connected - which affects exploration from both systems. Also, the last unlinked jump point can't connect to an existing system.

So 50% of the time, any system from 1-1000 will be chosen. Early on, that means a very low chance of connection. The other 50% of the time, it will be a system within a 30-system-number range centered on the existing system, but again unless those numbers contain existing systems there won't be a connection. Of course, every time you don't get a local chance hit, you move to a new area with none of the 30 local spread slots filled in. As you are moving around 50% of the time, most of the time you are in fairly barren local spreads. In fact, the second system within any local spread won't connect to the first one - because the connection already exists, so you are more likely to be in a new local spread before the current one can connect. Plus every time you hit a new system, the 'local spread' moves to include new unexplored slots. Even after 150 systems, you have only filled in 15% of the total slots, so the result you have is unlikely but possible.

For known systems, there is a fixed chance of finding a connection, but in random systems the total number of systems plays a large part. With 1000 systems, the chance of randomly hitting one of the filled-in slots is pretty low.

I just ran a similar experiment to yours, tracking all the generated system numbers and stepping through the code. I didn't find any links in 45 systems. Changed to 100% local chance (so I am not jumping around the whole 1000 systems) and hit a connection within 4 attempts.

I'm familiar with the math.

Attached below, the same experiment continued onward up to 302 systems. Still no loops or dead-ends. The map layout is different than the previous simply due to having to fit an additional 152 systems. I shall continue on and do it up to 500 systems later today, if I have the time and if it's necessary.

I just ran a new random stars map using 90 Local Chance and 10 Spread. There is a loop after three systems. Are you running any mods, or have you edited the DB at all? Also, its very odd (or very lucky) that you have generated 300 systems without running into the bug that causes a freeze in random stars games in v2.1.1. What version are you running?

BTW I'm not sure what you are asking me to do, but I am plainly not going to spend a lot of time chasing a bug that doesn't exist when I run the game.

[Ragnarsson]

I just ran a new random stars map using 90 Local Chance and 10 Spread. There is a loop after three systems. Are you running any mods, or have you edited the DB at all? Also, its very odd (or very lucky) that you have generated 300 systems without running into the bug that causes a freeze in random stars games in v2.1.1. What version are you running?

BTW I'm not sure what you are asking me to do, but I am plainly not going to spend a lot of time chasing a bug that doesn't exist when I run the game.

The database is the latest version, freshly unzipped, as is the .exe. Nothing has been modded or in any other way altered. Both the .exe and the .db are the latest versions available.

I have run into the bug that causes freezes upon system generation - many times. I have saved approximately every 10 systems I generated, and when the bug occurred I killed the process through the task manager and continued onward from the point at which I saved. Irritating, but able to be worked around.

If I were to speculate, and I'm no programmer so take that for what it's worth, it's something to do with the default settings (50% chance local / 15 spread) that's causing the problem. Loops can be generated with excessively high % chance, as you've shown, but the lower chances seem not to. I couldn't even begin to guess as to why.
(EDIT: Or the "avoidance of closed universe" changes that came in v2.1.1, as the issue does not exist when using v2.1.0).

I'm not asking you to do anything, this is your project and I'm simply an enjoyer of it. I am reporting an issue that exists when I run the game, an issue that has been encountered by many people as reported in this bug thread.

[db48x]

EDIT - this situation was exacerbated by another problem. I just realised that I was using Sol Diameters for the size of Stars, when everything else is in AU. That meant for the purposes of the above calculation, the stars were assumed to be 100x larger than their actual diameter, which lead to the min-max problem happening far more often. Aaargh! This went unnoticed for so long because I don't play random stars games.

I love good bug stories like this. Thanks for sharing!

[paolot]

During a travel though some jump points, is it possible that a ship class increases its mass? an increase that prevents the next jump(s)?
...

Were the other jump points stabilised? The ship should not be able to jump otherwise.

Really thank you Steve of the response.
No, the other jump points were not stabilised. I prefer each ship can travel on its feet and I don't use JP stabilisation.
Unfortunately, I cannot help you as I can't recall the DB of the game I was playing at that time, to post here. Sorry!

[Steve Walmsley (OP)]

I just ran a new random stars map using 90 Local Chance and 10 Spread. There is a loop after three systems. Are you running any mods, or have you edited the DB at all? Also, its very odd (or very lucky) that you have generated 300 systems without running into the bug that causes a freeze in random stars games in v2.1.1. What version are you running?

BTW I'm not sure what you are asking me to do, but I am plainly not going to spend a lot of time chasing a bug that doesn't exist when I run the game.

The database is the latest version, freshly unzipped, as is the .exe. Nothing has been modded or in any other way altered. Both the .exe and the .db are the latest versions available.

I have run into the bug that causes freezes upon system generation - many times. I have saved approximately every 10 systems I generated, and when the bug occurred I killed the process through the task manager and continued onward from the point at which I saved. Irritating, but able to be worked around.

If I were to speculate, and I'm no programmer so take that for what it's worth, it's something to do with the default settings (50% chance local / 15 spread) that's causing the problem. Loops can be generated with excessively high % chance, as you've shown, but the lower chances seem not to. I couldn't even begin to guess as to why.
(EDIT: Or the "avoidance of closed universe" changes that came in v2.1.1, as the issue does not exist when using v2.1.0).

I'm not asking you to do anything, this is your project and I'm simply an enjoyer of it. I am reporting an issue that exists when I run the game, an issue that has been encountered by many people as reported in this bug thread.

I've explained above why the 50% chance doesn't really generate local systems, so you are effectively just generating random numbers from 1000. Higher chances have a much greater effect. However, all that number does (50 or 90) is change the random chance - it doesn't affect how the game runs or how systems are generated. The actual code that references it is shown below. All changing the LocalChance number does is change the chance that the associated piece of code runs.

                    if (LocalChance > 0 && LocalSpread > 0)
                    {
                        if (GlobalValues.RandomNumber(100) <= LocalChance)
                        {
                            if (GlobalValues.RandomNumber(2) == 1)
                                DestinationNumber = StartSystem.SystemNumber + GlobalValues.RandomNumber(LocalSpread);
                            else
                                DestinationNumber = StartSystem.SystemNumber - GlobalValues.RandomNumber(LocalSpread);
                        }
                        else
                            DestinationNumber = GlobalValues.RandomNumber(NumSystems + 1) - 1;
                    }
                    else
                        DestinationNumber = GlobalValues.RandomNumber(NumSystems + 1) - 1;

Even given the above, it does seem odd that you have not generated any links in 300 systems in a clean game. The only other explanation I can envision is that there was a bug, but at some point I fixed it. Although I don't use Random Stars and I haven't posted any bug fixes to that effect, so maybe I fixed it by accident when looking at something else. Could you try using 90% local chance and see what happens?

Does anyone else use random stars and are they generating loops? I accept that a few people may have no loops, but unless no one at all is getting loops it doesn't seem likely to be code-related

I think one conclusion from this is that the default Local Chance should be higher, but that is probability rather than a code issue.

[Ragnarsson]

Could you try using 90% local chance and see what happens?

Happy to. Attached is the map resulting from that test.

- 90% local, Spread of 10
- 39 systems explored.
- System numbers  992,993,996-999,2-8,10-15,17-21,23,24,27 along with Sol of course and a smattering of non-local systems which don't seem relevant to this test, but can be seen on the map image. No dead-ends, no loops.

I suppose it's possible it was already fixed via some other change or fix, and I'm perfectly happy to just re-test after 2.2.0 is released to see if the issue persists. It makes no sense to waste time on something you can't replicate.

[Steve Walmsley (OP)]

Could you try using 90% local chance and see what happens?

Happy to. Attached is the map resulting from that test.

- 90% local, Spread of 10
- 39 systems explored.
- System numbers  992,993,996-999,2-8,10-15,17-21,23,24,27 along with Sol of course and a smattering of non-local systems which don't seem relevant to this test, but can be seen on the map image. No dead-ends, no loops.

I suppose it's possible it was already fixed via some other change or fix, and I'm perfectly happy to just re-test after 2.2.0 is released to see if the issue persists. It makes no sense to waste time on something you can't replicate.

Thanks for the testing. Although I can't prove it, I think the most logical conclusion is that a bug existed in v2.1.1, but sometime between the release of v2.1.1 and now I accidentally fixed it (and didn't realise).

[nuclearslurpee]

I've done a quick test with 90/5 settings, and I think I have worked out what is probably causing the observed behavior (see attached map):

I generated about 25 systems before getting a loop, which was from System #4 to System #1. This loop was only generated after all systems with IDs within ±5 had been generated (999, 1000, 1, 2, 3, 5, 6, 7, 8, 9). My guess is that this is probably related to the change to prevent dead-end star maps, and I agree with Steve that he probably fixed it accidentally as there were a couple of bug reports which I recall were found to be related to that new code.

Steve reported getting a loop within 4 connections at 100% local chance, which imples the bug was fixed unless the settings were something like 100/2, so I would guess it is indeed fixed by accident.


----

Side note: BUG REPORT
I got error #1654 in some cases when generating systems while doing these tests. Not sure if this is unique to Random Stars, but I don't recall seeing it in Real Stars games. I have all spoilers off, so it's not the error that happens when Rakhas generate sometimes.

[Steve Walmsley (OP)]

Side note: BUG REPORT
I got error #1654 in some cases when generating systems while doing these tests. Not sure if this is unique to Random Stars, but I don't recall seeing it in Real Stars games. I have all spoilers off, so it's not the error that happens when Rakhas generate sometimes.

That is the CreateNewRace code. It could be related to the creation of shipping lines for pre-industrial populations, as I fixed that now. Was there any NPR is the affected systems?

[nuclearslurpee]

Was there any NPR is the affected systems?

I did not check at the time, but the DB shows a pre-industrial race generated in that game so I think you've got the right of it.

[ExChairman]

I have had a loong game that while I saved last time, started throwing up many error messages, don't remember now, but almost all my systems jumppoint information is lost, well that would have been a fun "happening", rediscover the empire... Perhaps making them free and having to conquer again..., but you cant use the existing jumppointinformation, getting error #3102,  need to regen the systems jumpoints and then its new jumppoints will go to new systems... Most info about the game is still there... Trying the fleet window, error "" then moving time, getting error #3476, followed by# 700, #697, #3312, restaring the error string...

Well so sad, the game is busted...

[Noriad]

A few months ago I started a new game. Version 2.1.1. Max systems 80, Known Star Systems, 3 starting NPRs, conventional start. Known Star Systems selected because I like to restart a few times until I have a good starting position and without Known Star Systems selected, half the times during generation the game would freeze forever.
 
I decided that 80 systems is the most micromanagement I could tolerate, 3 NPRs (non-player races) would be a decent challenge (I assume), and if I would manage to effectively conquer all 80 systems, I would consider the game won.

However, after a few weeks of playing, I have discovered 81 systems, with a few dozen unexplored jump points, and aside from a few "spoilers", I have not made any contact with any NPR. So my desired setup failed. And unlike earlier versions of the game, no cross-links between system branches were found on the Galactic Map. The whole Galactic Map was just star connections branching out further and further.

The 80 system limit clearly doesn't work, and I'm unable to find the NPRs. So I considered the game broken and quit playing the current version.

SJW: The system limit doesn't apply to Known Stars games, as per the associated tool-tip. There are 4500 systems in Known Stars games, with NPR proximity to Earth set in game options.

[nuclearslurpee]

Known Star Systems
...
The 80 system limit clearly doesn't work, and I'm unable to find the NPRs. So I considered the game broken and quit playing the current version.

The limit on number of systems only works for Random Stars games, this is clearly stated on the tooltip when you set up the game options. Not a bug. The way to get the desired effect is to turn off the Known Stars option once you have about the desired number of systems - at least, this should work, if it doesn't then there is a bug.

[Noriad]

Known Star Systems
...
The 80 system limit clearly doesn't work, and I'm unable to find the NPRs. So I considered the game broken and quit playing the current version.

The limit on number of systems only works for Random Stars games, this is clearly stated on the tooltip when you set up the game options. Not a bug. The way to get the desired effect is to turn off the Known Stars option once you have about the desired number of systems - at least, this should work, if it doesn't then there is a bug.

Fair enough. But that still leaves my game without NPRs.

[nuclearslurpee]

Most likely there are NPRs, you just have not run into them yet. Depending on your settings at the start of the game, it can be somewhat rare to encounter NPRs until you've explored quite a lot of systems - and other times they may show up on your front door. It's quite random.