Single Player bugs and crashes - After the 16th of May

Could you please point me where i can find it on my comp? Haven't done this in ages ...

Don't worry - I was able to reproduce it...more later when I have time to look into it in detail...

Edit - well this one's fascinating! To quote Scott 'I may be gone for some time...'
 
Don't worry - I was able to reproduce it...more later when I have time to look into it in detail...

Edit - well this one's fascinating! To quote Scott 'I may be gone for some time...'

=)

Guys, i seriously wish today's game producing companies had approach like 5% of yours (this mod team) to games they are making and selling us.

Really really sick of tons of crap they call "games" they throw at us gamers last years, be it MMO, RPG or whatever, with shooters being probably only exceptionm but i don't play them. It's like 1 games per 2 years that are worth buying. They are not mostly worth time even downloading with torrents. It's nowadays like this: make a trailer with blazing graphics (and tits ofc) - market and advertise it, sell boxes, PROFIT! and nobody cares that game die within 2-3 months. Zero new ideas, zero replayability, dumbed-down gameplay, corridor-type content "Press X to win" style mixed with few cutscenes.

This mod (with RoM and AND earlier) and couple other semi indy or modded games are ones i spend 95% my gaming time at for last 4 years or so.

Sorry for my bitter offtopic rant ... im just really sick and sad because of this.

p.s. Minidump file was 0 bytes.
 
=)

Guys, i seriously wish today's game producing companies had approach like 5% of yours (this mod team) to games they are making and selling us.

Really really sick of tons of crap they call "games" they throw at us gamers last years, be it MMO, RPG or whatever, with shooters being probably only exceptionm but i don't play them. It's like 1 games per 2 years that are worth buying. They are not mostly worth time even downloading with torrents. It's nowadays like this: make a trailer with blazing graphics (and tits ofc) - market and advertise it, sell boxes, PROFIT! and nobody cares that game die within 2-3 months. Zero new ideas, zero replayability, dumbed-down gameplay, corridor-type content "Press X to win" style mixed with few cutscenes.

This mod (with RoM and AND earlier) and couple other semi indy or modded games are ones i spend 95% my gaming time at for last 4 years or so.

Sorry for my bitter offtopic rant ... im just really sick and sad because of this.

p.s. Minidump file was 0 bytes.

Yeh, minidump is 0 bytes because the failure occurs when the process runs totally out of memory (no even enough to process the minidump). The reason I runs out is that it's trying to process a RIDICULOUS path (like crazily long and intricate) and in doing so discovers lots of small improvements causing it to repeatedly throw away large section of the path tree and recalculate (slightly) better variants). Because of the way memory allocation works within a single path generation (it never frees anything so as to have a very fast forward-only allocator on chained pre-allocated super-blocks) this means it eats more and more memory and actually turns out before it finishes processing this mega-path.

Arguably there are 3 problems here:

1) Perhaps the allocator needs to garbage collect within gentian of a single path in exceptional cases (highly complex to do - not going there currently)

2) Perhaps the path generation should accept decent, but not necessarily totally optimal paths when asked for the best route between very distant tiles. I'm considering this approach

3) Asking to generate such long paths is probably not a good idea anyway (lots of wasted time) in almost all cases. In this case an attack stack that is miles from any home area, REALLY deep in foreign territory, has fallen below critical size for staging a city attack, and is seeking to return to a staging city. That's not really sensible when the nearest city is so far away across hostile territory, so it should probably not bother. Honing the context in which it tries to return to a staging city (and thus not needing to generate the problem paths) is the approach I am working o now, and will probably be where we end up for V32 (it should improve performance too on large maps). I will probably implement (2) as well sometime after release as a fallback.

BTW - your map is really IMPRESSIVELY large! It's not a standard size is it?
 
Yeh, minidump is 0 bytes because the failure occurs when the process runs totally out of memory (no even enough to process the minidump). The reason I runs out is that it's trying to process a RIDICULOUS path (like crazily long and intricate) and in doing so discovers lots of small improvements causing it to repeatedly throw away large section of the path tree and recalculate (slightly) better variants). Because of the way memory allocation works within a single path generation (it never frees anything so as to have a very fast forward-only allocator on chained pre-allocated super-blocks) this means it eats more and more memory and actually turns out before it finishes processing this mega-path.

Arguably there are 3 problems here:

1) Perhaps the allocator needs to garbage collect within gentian of a single path in exceptional cases (highly complex to do - not going there currently)

2) Perhaps the path generation should accept decent, but not necessarily totally optimal paths when asked for the best route between very distant tiles. I'm considering this approach

3) Asking to generate such long paths is probably not a good idea anyway (lots of wasted time) in almost all cases. In this case an attack stack that is miles from any home area, REALLY deep in foreign territory, has fallen below critical size for staging a city attack, and is seeking to return to a staging city. That's not really sensible when the nearest city is so far away across hostile territory, so it should probably not bother. Honing the context in which it tries to return to a staging city (and thus not needing to generate the problem paths) is the approach I am working o now, and will probably be where we end up for V32 (it should improve performance too on large maps). I will probably implement (2) as well sometime after release as a fallback.

BTW - your map is really IMPRESSIVELY large! It's not a standard size is it?

It looks like Immense to me (300 by 200 tiles).
 
@Joseph - is the latest SVN behaving ok for you (i.e. - solved your end of turn crashes when the pipeline threads value is 4 or more)?
 
Yes it has, been playing with setting of 4 for over 20 turns.

JosEPh :)
 
Guys, i seriously wish today's game producing companies had approach like 5% of yours (this mod team) to games they are making and selling us.

Yeah I hear ya. The only game I am looking forwrard to is the Indie game, Starbound. i have been following their development for a long time now and can't wait to see it finished.

Another good Indie game I found last month was Reus. Its a stylized God game that has a lots of puzzles. I was also excited about the new Sim City, but my experience was heavily tainted by the release which was so buggy.

Now as for old games I still play Sims 2 with lots of custom content. Note I still love C2C and plan on working on it for a long time, but sometimes one needs a break since this has become both work and play for me.

On the upside those types of games have helped inspire me for new content. Ex. Marsh Mallows (I did not know such a plant existed) from Reus and I have even taken ideas for other games I don't even have like Anno 2070. Their wiki is great for Trans-Human ideas.

And if you could not tell most of the games I like are Sandbox/Builder style games. :goodjob:
 
Got a repeating build problem with the Alter of Human Sacrifice? I built it (pic 1) and then in the same instant it says i can build it again, and next turn it is there also??

I've found the bug that causes this (fix will be on the SVN later today), bu it raises a design point that DH might want to think about:

@DH - the situation is this:

1) 'Alter for Human Sacrifice' is Replaced by 'Sacrifical Alter'
2) 'Sacrifical Alter' goes obsolete with tech 'Humanism'. However 'Alter for Human Sacrifice' has no obsoletion tech

The net effect is that in cities with a sacrificial alter (SO's capital had one), you cannot build an alter for human sacrifice (so far so good) UNTIL you research Humanism! At that point the sacrificial alter goes obsolete, and you can build the alter for human sacrifice again (the bug is that when you build it, it does not show up in this situation, but I'll fix that).

So basically Humanism is ENABLING the alters for human sacrifice! Perhaps Humanism should also obsolete the entire worldview on which this alter rests??
 
I thought Inniskilling was introduced as Irish culture unit but it doesn't require Irish?
 
Something wrong with the AI when i use the prophets option. The Ai lost the interest in religions when dont be the first to research the religion. Well, the ai can found the religion if develop and have a prophet to fund the religion, so dont need to lost interest, just change the strategy.
 
Something wrong with the AI when i use the prophets option. The Ai lost the interest in religions when dont be the first to research the religion. Well, the ai can found the religion if develop and have a prophet to fund the religion, so dont need to lost interest, just change the strategy.

I take it this is on a non-choose religion game?

I don't think much ai work has been done on the tech research decisions when under DP so yeah, little issues like that may remain. It wasn't an issue until they were made to be worthless techs if not seeking that specific religion.

Can't promise an immediate fix here but I can put it 'on the list'. Thanks for taking the time to note it.
 
Top Bottom