PROJECT: Playtest Thread III

[Ekmek]

Don't fret Tamerlin, as my threads prove I'm only learning now about programing But I think there is something in the code that calculates turns until growth (maybe citydata.cpp) if you figure out how to display that I think you'll be on the right track...

[J Bytheway]

Martin: Have your recent attempts to fix the AI-slowness problem had enough effect to make it worthwhile releasing a new playtest version? If so, I'll throw one together.

[Martin Gühmann]

Originally posted by J Bytheway
Martin: Have your recent attempts to fix the AI-slowness problem had enough effect to make it worthwhile releasing a new playtest version? If so, I'll throw one together.

I think this attempt is effective enough. However there is another problem that makes the game crash concerning the changes I did about vision. I was distracted by some other stuff like the AI specialists distribution. And now I am trying to figure out why the AI doesn't build special units and settles so slowly.

I plugged in the build lists of GoodMod, and the results were that the AI settles agressively, but doesn't build any wonders. Now I want to look into the change about reducing the city garrision. That might have an effect as there is a GarrisonUnitBuildList flag for a BuildListElement.

However without this side stuff I would have released a playtest myself.

-Martin

[Martin Gühmann]

Well it is time for a new playtest. It is again a .NET build so these these instructions are still valid:

Do I need any additional files to make the playtest work?

It is a build created with the .NET compiler. Therefore you need the .NET runtime environment to run the *.exe.
If you dont already have it it is avalaible at windows update or here.
Or at least you need those two dlls if you don't want to install the .NET runtime.

I was able to reduce the massive slow down problem in more than one continent games. However it seems not gone totally. But how severe the problem is can only show a big playtest.

Here is the change log since the last playtest:

2005-03-05
Disabled: AI unit text, so that you don't see the AI goals.
Fixed: Crashes and memory leaks.
Fixed: Minor translation problems with the French and German string files.
Updated: strategies.txt, advancelists.txt, buildlistsequences.txt and unitbuildlist.txt,
to make the AI more agressive and seperate the AI personalities more.
Fixed: Vision of borders of civs you have no contact to in fog off and god mode.
Fixed: A Vision problem and a crash in combination with city conquest.
Changed: Disabled autosave for MP client.
Added: Finished production handling in the MP turn start.
Updated: Anet game number.
Fixed: Basic AI population assignment.
- If cities has slaves the number of pops to assign is calculated correctly.
- The first pop in the database can now be used, meaning AI can use entertainers.
- The minimum amount of calculated entertainers can now be assigned.
- However this are only basic fixes, population optimisation has still to be done.
Added: A new slic power operator: **
Fixed: Slic database access, so that the feat units in CTC work.
Added: Slic function AddSlaves to add slaves to a city.
Added: Slic function IsOnSameContinent to check whether two locations are on the
same continent.
Fixed: Wonder start and almost complete messages are also send for the first wonder
in the database correctly. Precisely you get now such messages for Aristotle's Lyceum.
Fixed: Some minor bugs in the AI goal computation.
Restored: Old operator of squd strengh. That caused the huge delays in the last version
on maps with more than one continent.
Changed: Pathfinding doesn't consider anymore danger on the way. Another solution for
that has to be found. This removes the huge zone of control.
Changed: Tile improvements for small cities of AIs get full utilisation bonus for placement.
Changed: AI places tile improvements every fifth turn and roads every second turn.
Added: CanSettleOn flag to unit.txt so that modders have the ability to limit settling
to certain terrains.
Changed: New games can now be started with only one civ (without any opponents).
Changed: The version date is now derived from the exe's last link date (date is in GMT),
to prevent showing a wrong date if ldl_str.txt has not been updated.
Fixed: Turn counter synchronisation when creating a new civilisation (by revolt) to
prevent one sided wars.
Fixed: Science popup in hotseat and PBEM is delayed until the appropiate player's turn.
Fixed: The risk levels in the drop down menu in the scenario editor are now in the
right order.
Fixed: Turn count in PBEM don't say anymore always "Turn 0"
Changed: The same algorithm for SP and MP is used to generate Barbarians
Changed: New terrain values

And finally a link to the latest playtest:

Edit: Link removed, there is already a new version out. Check later posts of this thread thread or the first post.

-Martin

[Fromafar]

Changed: The version date is now derived from the exe's creation date (not last modification date)

Did you change something here, or did I mess up something in the code? It should really be the last modification (= link) date of the file. The file creation date is irrelevant.

[Ekmek]

I think this might be the first bug with te new playtest, and I'm partly to blame

I tried the new CanSettleOn flag. So I went to the text file and deleted all of the Settle: Land flags and put in CanSettleOn: Terrain_Grassland. The FLAG works !!!!!!!

HOWEVER, when I started the game it came up as the urban planner AND NOT the settler you start with. The way I knew my flag worked was that I had to go to the cheat mode and then add settlers (after THAT i tested it out)

I also went back into the units.txt and made it only CanSettleOn and just Settle: Land. ANd it settled any land (overrided the flag) and it started with the right settler.

So I think there is a bug or code somewhere that happens during start up to check for the best starting settler? But it doesnt make sense because the tech wasn't available...I'll look in the code but not sure where to start...

[The Big Mc]

Marin that’s not my addslave function is it couse it has a bug.

[Ekmek]

Still looking but I think it has something to do with the Settle: flag. I think the game checks for those flags to see if a settler and then picks the first available one. I'll try to find the check and add CanSettleOn to it.

BigMC whats your bug? also I played with my tiles briefly and copied volcanoes but made them land and changed the tile set and came out with single square islands (no beach tiles) and it appeared on start up so I think there is something about 'internal' for the game to identify terrains which makes it hard coded to only use the ones on startup and not new ones we add.

[Maquiladora]

When I bombard a city on one turn, on the next turns it automatically bombards for me, for example after many turns of bombarding I wanted to attack but after I move my other non-seiging units it switched to my previously bombarding stack and did it automatically.

There is a way to stop it by selecting and moving/using that bomarding stack first before any other units (I think turning off auto-cycle units might work too) so you catch it before it decided to autobombard.

Ive attched the save. Just move the 3-stack (archer, warrior, archer) closer to the orange AI's 2 settlers then it switches to the 12 stack and bombards by itself.

NOTE: and no Im not giving the command to bombard twice on the previous turn so it waits till the next turn and auto does it.

This is obviously with the new playtest.

[Maquiladora]

Also you can still tell if a city is empty by bombarding it, if its empty it wont allow it.

And the new government discovery popup still has the white text and missing name.

[sekel]

Originally posted by Maquiladora
Also you can still tell if a city is empty by bombarding it, if its empty it wont allow it.

I guess that the bomber´s crew flying over the city can notice it is empty

[Maquiladora]

Flying catapults?

[Maquiladora]

Originally posted by Maquiladora
NOTE: and no Im not giving the command to bombard twice on the previous turn so it waits till the next turn and auto does it.

Actually I just realised it is giving the bombard order twice, but its not me doing it.

I give the order to bombard, my units bombard but then the "Clear Orders" command is in the right-click commands menu, so its waiting to bombard again, which has to be next turn.

This also happens with incite revolution with the spy, it gives the order twice. I guess this happens with all orders.

[Martin Gühmann]

Originally posted by Fromafar
Did you change something here, or did I mess up something in the code? It should really be the last modification (= link) date of the file. The file creation date is irrelevant.

I didn't change anything. The point could be the link issure, clearly Windows Explorer says the file was last modified on the 5th. However I only modifyed the comments of the source files and ran the compiler. So the *.exe should be identical with the one I creted before midnight. However I cannot know for sure, whether the *.exe was relinked.

Originally posted by E
Still looking but I think it has something to do with the Settle: flag. I think the game checks for those flags to see if a settler and then picks the first available one. I'll try to find the check and add CanSettleOn to it.

Indeed it has to do something with one of these flags, of course not with your flag. But actual you should find the place in the code on your own.

Originally posted by Maquiladora
When I bombard a city on one turn, on the next turns it automatically bombards for me, for example after many turns of bombarding I wanted to attack but after I move my other non-seiging units it switched to my previously bombarding stack and did it automatically.

There is a way to stop it by selecting and moving/using that bomarding stack first before any other units (I think turning off auto-cycle units might work too) so you catch it before it decided to autobombard.

Ive attched the save. Just move the 3-stack (archer, warrior, archer) closer to the orange AI's 2 settlers then it switches to the 12 stack and bombards by itself.

NOTE: and no Im not giving the command to bombard twice on the previous turn so it waits till the next turn and auto does it.

This is obviously with the new playtest.

And this really occured in this playtest for the first time? My first thought is that it has to do with the modified bombard code. But that was already in the last playtest build from late December. However if there are all orders affected then it might be something else.

Originally posted by The Big Mc
Mar[t]in that�s not my addslave function is it couse it has a bug.

Well I added my version of this function, since I had to test it on my own. At testing it didn't crash the game, however the Assert poped up that the AI had problems with distributing its specialists. But that was a problem common to all slaves regardless how they were added to the city.

I realized that the problem was located something else. And left the function in. It has the advantage that it uses the MakePop event, so you can detect it when it occurs and do some stuff via slic whenever a slave is added to a city. However I didn't implement the remove slave part, but that shouldn't be a big deal.

-Martin

[Maquiladora]

And this really occured in this playtest for the first time? My first thought is that it has to do with the modified bombard code. But that was already in the last playtest build from late December. However if there are all orders affected then it might be something else.

Its definitely all orders, well all the ones ive tried so far, all except some of the special units, slave raid and pillage does repeat though.

It stops when you ungroup then group the stack before the next turn, of course all this does is clear orders so its the same thing.

[The Big Mc]

Has anybody seen the combo box in slic messages I noticed it when I was doing a bit of slic but every time I hit submit it crashed.

[Fromafar]

On the date issue: it is not as bad as I thought. The executable is correctly displaying the last modification date, but is using GMT as time zone.

So, we only need to update the change log text, and have to be careful when releasing between 0:00 and 1:00.

[Martin Gühmann]

Originally posted by Fromafar
On the date issue: it is not as bad as I thought. The executable is correctly displaying the last modification date, but is using GMT as time zone.

In this case it ok.

Originally posted by Fromafar
So, we only need to update the change log text, and have to be careful when releasing between 0:00 and 1:00.

Why updating the change log I released the playtest on the 5th in both time zones GMT and CET.

But another question to all, the game now in a state so that we can now advertise it? And if yes, who is going to write the news item for our civ group news.

-Martin

[Ekmek]

Yes, I'd say we need some advertising. I think we need to have a readme advertising all the work displayed too (cant recall if there was one with the new build and its not here at work)

[Martin Gühmann]

Originally posted by E
Yes, I'd say we need some advertising. I think we need to have a readme advertising all the work displayed too (cant recall if there was one with the new build and its not here at work)

Of course I updated the readme, but I would like to see here someone who tested the latest playtest writing a little news item for our civ group so that everyone knows at Apolyton that there is a new playtest.

-Martin

[Ekmek]

I haven't played a full game yet, but I think Maq has

[Maquiladora]

I havent finished, its upto 358 turns, but doesnt look like finishing soon, even if i try for a diplo-victory.

Ill write something anyway once i finish it, as a report, and news item if its not too negative.

But another question to all, the game now in a state so that we can now advertise it? And if yes, who is going to write the news item for our civ group news.

I would say its not in a state to advertise because of the long pauses (nothing more than civ3 though) late in the game and the repeating orders problem i mentioned. Perhaps the long pauses wouldnt be so bad if the sound of a moving unit in my vision didnt repeat while that civ was moving its other units elsewhere under the fog of war.

[Ekmek]

True, it has issues, but a little publicity might bring more testers an hopefully more programmers, let them know the project hasn't died. I'd recommend making the news item at Civfanatics too (even if its just a thread).

[Maquiladora]

There are plenty of positives but the negatives are really noticeable things, definitely more so than the improvements, especially to someone new to the playtest.

But of course it can serve as a general update and as you said show its still going and steps forward have been made, but not necessarily an advertisement for people looking for the complete civ game... yet.

[Fromafar]

Originally posted by Martin Gühmann
Why updating the change log I released the playtest on the 5th in both time zones GMT and CET.

Because the change log states that the file creation date is displayed, which is not true.

But another question to all, the game now in a state so that we can now advertise it?

Not really. As indicated by others, the delays are still too long. I think we should investigate the impact of the "subtask" changes by Calvitix, and also improve the <-operator to become anti-symmetric at least. I do not have much confidence in the result and speed of a sorting algorithm with the current operator.

[Martin Gühmann]

Originally posted by Fromafar
Because the change log states that the file creation date is displayed, which is not true.

I see and at least I have updated the version I have posted in this thread.

Originally posted by Fromafar
Not really. As indicated by others, the delays are still too long. I think we should investigate the impact of the "subtask" changes by Calvitix, and also improve the <-operator to become anti-symmetric at least. I do not have much confidence in the result and speed of a sorting algorithm with the current operator.

I suggested here a new version of the operator. However it seems to me that the speed of the sorting algorithm is not too badly influenced by that operator. So the subtask stuff and the other changed should be investigated. Unfortunatly can't find his name in every file he modified, so I overlooked some of his changes.

Originally posted by Maquiladora
Perhaps the long pauses wouldnt be so bad if the sound of a moving unit in my vision didnt repeat while that civ was moving its other units elsewhere under the fog of war.

Well this reminds me of this problem, maybe it is unrelated:
"Always terminate a unit's sound effects when it dies"

The other issure seems again be related to the AI transport problem. Indeed in my test game only those AIs are effected that have to transport a lot of units. At least they now move their units to another continent, but I have the feeling that they could better, fill their transports, unite their forces on the other continent before attack. And focus on one city. Well this would be also a good idea for cities on the same continent.

However I think we need a newsitem. But of course the question is how should it look like.

-Martin

[J Bytheway]

Originally posted by Martin Gühmann
I suggested here a new version of the operator. However it seems to me that the speed of the sorting algorithm is not too badly influenced by that operator. So the subtask stuff and the other changed should be investigated. Unfortunatly can't find his name in every file he modified, so I overlooked some of his changes.

Since I've put things into version control, I can easily provide you with diffs detailing exactly what he changed, and in which files. If that would help, just ask.

[Martin Gühmann]

Originally posted by J Bytheway
Since I've put things into version control, I can easily provide you with diffs detailing exactly what he changed, and in which files. If that would help, just ask.

That would be great unfortunatly I won't have the time to look into it for the next week, therefore you should post it in the Army Movement thread so that others have a chance to look into it as well.

-Martin

[Qumad Al-Riyf]

I downloaded the 2004.03.05.CTP2.Playtest.rar file and the .NET compiler.
I installed the .NET compiler and CTP2 worked fine.
Then I installed the playtest in the main CTP2 folder and when I try to rin the game I get this message:
"
CTP2.exe - Unable To Localate DLL

The dynamic library MSVCP70.dll could not be found in the specifid path
D:\Games\CALLTO~1\CTP2_P~1\CTP;.;C:\WINNT\system32 ;C:\WINNT\system;C:\WINNT;C:\WINNT\system32;C:\WIN NT;C:\WINNT\System32\Wbem.

[OK]
"

Originally posted by Martin Gühmann
Well it is time for a new playtest. It is again a .NET build so these these instructions are still valid:

quote:
Do I need any additional files to make the playtest work?

It is a build created with the .NET compiler. Therefore you need the .NET runtime environment to run the *.exe.
If you dont already have it it is avalaible at windows update or here.
Or at least you need those two dlls if you don't want to install the .NET runtime.

When folowing that link I get to a page that tells me the following:
"No input file specified."

So that dident work ether, any idea?

[Solver]

Martin made a mistake with his link. The following link works for me. Unzip the two files into your system folder.

http://apolyton.net/forums/attachmen...postid=3466464