gae 0.2.13-development

[hailstone]

In Glest 3.2.2 the arrow doesn't appear at all when right clicking on a unit/building. I haven't figured out how this is done though.

Its an error which is marked in line 186 in file game/types/flags.h

The message doesn't say what's wrong with compiling. Looks like it has been marked as a breakpoint for debugging.

[jda]

The red arrow is useful when you're attacking a position, not an unit - you know, when you send the unit in attack mode to an empty tile and it will attack any foe it finds along the way.
It is pretty pointless when applied to an unit though.

---
hailstone posted just before I clicked to post the above. I must say: What are you talking about ???

[silnarm]

For the red arrow problem: I'm only talking about the red arrow when ATTACKING a unit! In general this is a good arrrow that helps playing. I would only remove it in the case of a an attack command.

Yes, if a unit is clicked I think there should be no arrow, the code that places them is in Gui::giveDefaultOrders() and Gui::GiveTwoClickOrders(),
change

Code: [Select]

mouse3d.show(targetPos);to,

Code: [Select]

if (!targetUnit) mouse3d.show(targetPos);
Which should give us the desired behaviour, I'll can test and get this in tonight (err, that's +6/7 hours from now ) if someone wants to try it out a beat me to it, please do!

For the particle problem found by softcoder:
I don't have these problems here with gae ( latest 0.2.x )

The picture Eliminator posted looked very familiar, so at least one other person had the same problem.  I'm quiet happy softcoder had this problem, as he seems to have fixed a problem that I couldn't reproduce.  I'll try the new code with Omega's Military tonight, I think that's the only complete mod the uses random particles.

Anyone has ideas for the following?
Its an error which is marked in line 186 in file game/types/flags.h

Some quick googling indicates this is an eclipse thing, when it get multiple error on the same line it just gives that message, rather than all the actual errors... not very helpful.

Could you please run make from the command line and capture the actual errors and post them, something along the lines of make 2>&1 | grep 'error:' (i think that's right for bash).

[softcoder]

Testing with latest 0.2.x codebase thus far:

Immediately notice that single player (so far) seems more stable (compared to 3.2.2_network). Also noticed that the particle problem STILL IS THERE with 0.2.x (even though Titi says he doesn't see that particle problem? Could it be that Titi has changed code due to his particle changes?)

EDIT:

I mention the particle problem cause it I thought maybe it was not a problem in the 0.2.x branch since titu didn't seem to see it, but right away we see it with a build from 0.2.x so we may need to apply the fix I mentioned earlier?

Also noticed that saving and re-loading a saved game (0.2.x codebase) does not work.. units seem "stuck" when reloading a saved game and cannot move.

So assuming the non network play is stable, should I port the network changes to 0.2.x branch? Which branch is stable enough to work in?

Thanks

[silnarm]

Immediately notice that single player (so far) seems more stable (compared to 3.2.2_network). Also noticed that the particle problem STILL IS THERE with 0.2.x (even though Titi says he doesn't see that particle problem? Could it be that Titi has changed code due to his particle changes?)

No. This problem affects only some linuxes, when it was first reported some time ago I could run the game under linux and was not getting the same problem, hence it never got looked at.

So assuming the non network play is stable, should I port the network changes to 0.2.x branch? Which branch is stable enough to work in?

There is very little difference between 0.2.x and 3.2.2_network, the net branch was only created recently and has been synchronised with 0.2.x twice since, the vast majority of the differences are in the networking code, and the only other changes were minor, and unlikely to be the source of the problem.

Could someone who can run the game under linux please confirm that an unmodified build of 3.2.2_network is seg-faulting ??

Cheers

[softcoder]

I'm checking out 3.2.2_network fresh and getting a build and we'll test it.

The boys noticed some weird behaviour with the pathfinder (using the 0.2.x branch).

They say that units were getting "stuck" in a running motion but not moving and the animation of a crumbled building did not disappear after a while, it just stayed there all game. I think right now the biggest noticable issue is the pathfinder (for 0.2.x) as units seem to get very confused and unable to carry out commands from time to time. If required i'll get the boys to take screen shots and explain in more detail.

[titi]

Same bugs here, I have screenshots! ( 0.2.x branch )

Other bugs:
- Autorepair mode. Units sometimes stop their commands to repair something!
I told a unit to build an energy tower, suddenly it turned around and repaired another. I also saw a gold minig initiate suddenly start repairing a building. I think this whole autrepair mode is a bit buggy.

- I saw a horserider( cpu controlled ) running to the battle. When he arrived he started to walk on place ( like the other bug ) instead of attacking

Here hte screenshots of the other bugs:
first the running in place man, second the building that does not disappear. ( These running units which get stuck happen very very often when there are lots of units in the game.)

[-Archmage-]

Actually, I think it would be cool if what you last saw on whatever part of the map would stay that way until you see it again!

[titi]

Ok a second post for something more positive  
Mega-CPU and Easy-CPU work in GAE now.
I made the changes in 0.2.x branch. What should i do with the code? checkin? Send it to someone?

[silnarm]

- Autorepair mode. Units sometimes stop their commands to repair something!
I told a unit to build an energy tower, suddenly it turned around and repaired another. I also saw a gold minig initiate suddenly start repairing a building. I think this whole autrepair mode is a bit buggy.

Ok, I disabled auto-repair for AI units to prevent the problems with idle workers, I think for now we may have to disable it for everyone until we can sort out what's going on here.

re the pathfinder problems, I am aware and can reproduce, but I'll have to ask for a little patience here, it's very difficult to figure out what is causing such problems, and when the behaviour only manifests itself with a lot of 'active' units, my fancy textures and overlays don't help a lot... I will find it, but it will require some creative logging and may take a little while to sort out.

Actually, I think it would be cool if what you last saw on whatever part of the map would stay that way until you see it again!

This is indeed how it would ideally work, and hopefully one day it will...

Ok a second post for something more positive 
Mega-CPU and Easy-CPU work in GAE now.
I made the changes in 0.2.x branch. What should i do with the code? checkin? Send it to someone?

Yay! Something positive!

Ok, exactly what you want to do here will depend on how comfortable with SVN you are... from the screenshots you posted I see you still have the water rendering problem at the screen edges, so your working copy is not up to date, and if you try to commit now, it will fail.

My recommendation would be,

1. Create a patch, just in case the following messes up your working copy.
Issue a command like this : [ svn diff > ai_players.patch ] in the top level dir of the checkout.

2. svn update, and then...

If you get no conflicts, rebuild and ensure everything is ok (it should be, but test it anyway), then commit.

If you get conflicts and you are comfortable enough with SVN to resolve them, go ahead and resolve them, then test, and commit.

If you get conflicts and you are not comfortable enough with SVN to try and resolve them, send the patch you created to me, and I'll merge it for you

Of course, if you're really not sure about all this conflict business, feel free to perform only step one and send it to me

[softcoder]

More positive results:

The boys report back that the 3.2.2_network branch is NOT unstable, so that leads me to believe somehow when I ported the network code it affected stability (even of the single player experience). I am carefully diffing my branch to the vanilla 3.2.2_network branch to see what is wrong. My apologies for assuming it was GAE, it appears that it is not GAE causing the issue. Once I get things under control I'll discuss further.

The strange part is that the same code in mega-glest was stable, so I'm not sure what is wrong, but I'll find it.

Thanks

EDIT:

wrong... there appears to be things bad in the 3.2.2_network branch, some random and some not so random. The auto-repair seems to crash every time in a specific scenario (not sure it it happens in all scenarios).

I'm going to port the network changes to my copy of the 0.2.x branch and try that out.

[hailstone]

Perhaps requestCommand() has something to do with the problems since it's in 0.2.x but not 3.2.2_network.

Code: [Select]

void ClientInterface::requestCommand(Command *command) {
if(!command->isAuto()) {
requestedCommands.push_back(new Command(*command));
}
pendingCommands.push_back(command);
}
Keep in mind that 0.2.x is meant to be a bug fix release to get GAE working with all the 3.2.2 features. Any new features should really go in trunk for 0.3 (but if it's ok with silnarm I don't mind; it would be good to get in the habit though). The save game feature is known to have bugs and since it's not a feature of 3.2.2 I was thinking about disabling it until 0.3 so a stable release can be out sooner.

[silnarm]

Keep in mind that 0.2.x is meant to be a bug fix release to get GAE working with all the 3.2.2 features. Any new features should really go in trunk for 0.3 (but if it's ok with silnarm I don't mind; it would be good to get in the habit though).

Yeah, I'm happy for things that are known to be working to be added (ie, titi's AI enhancements) but otherwise the focus should definitely be on getting known issues sorted out.

The save game feature is known to have bugs and since it's not a feature of 3.2.2 I was thinking about disabling it until 0.3 so a stable release can be out sooner.

I agree, at this stage I think anything not in 3.2.2 that is currently causing us headaches should be 'switched off' for now.

Re: Pathfinder
I've committed a fix, it stops the units running on the spot, but the underlying issue still needs resolving, but it has at least been identified now.

Re: Particles
I also cleaned up the particle stuff (completely removed sPlayLast), so sorry if I've messed things up for you softcoder, hopefully you'll be able to update your working copy without too many problems, and if so, please let me know if the particle fix was a success.

Cheers.

[titi]

I checked out( without conflicts) and I'm ready to checkin, but there is a new error!!!
I played in map mountains with tech. Games starts and I start to walk around with a worker. This worker uddenly slows down or speeds up ???
What is this?  Is someone trying to speedup or slowdown units when walking up or down a hill and made some bugs?? And I really don't like this feature yet, because this requires major changes in the pathfinder for example!

Should I checkin nevertheless?

[silnarm]

I checked out( without conflicts) and I'm ready to checkin, but there is a new error!!!
I played in map mountains with tech. Games starts and I start to walk around with a worker. This worker uddenly slows down or speeds up ???
What is this?  Is someone trying to speedup or slowdown units when walking up or down a hill and made some bugs?? And I really don't like this feature yet, because this requires major changes in the pathfinder for example!

Should I checkin nevertheless?

Yes, but you may have to update again first   I committed a small fix for the ScriptManager a couple of hours ago, it should update without problem.

The 'height factor' on move speed had indeed been radically altered from the value used in vanilla Glest, I'm surprised I never noticed that, it is indeed very obvious and not so nice on the map Mountains.  I've restored it to the value from 3.2.2, but I'll hold off committing for now so you can get your changes in without futher issues.

[titi]

ok its in cvs now

Update: well .... svn of course 

[silnarm]

ok its in cvs now

Thanks titi!

I've committed the move speed fix, and a re-fix for minimap lag.

[silnarm]

... so ... very ... tired...

I have managed to reproduce the seg-fault in 3.2.2_network and have it identified, command archetypes are no longer being handled correctly, cancelling a command for example is a sure fire way to crash it.

I have also identified the sync problem, attacks are not panning out in the same order, probably because of auto-commands, and probably related to the 'missing code snippet' hailstone pointed out earlier.

So, if you haven't given up on us already softcoder, please bear with me, my head is hurting a little after meticulously pouring over and comparing a pair of 43Kb logfiles... I'll get it sorted... maybe tonight, but more likely tomorrow sometime (Australian eastern time, GMT +11).

Cheers.

[-Archmage-]

Talk about tired, it's 3:30 in the morning where I live!
Although, I'm kinda nocturnal, so I'm not very tired...........well, lets see me get up...............maybe later......................
My Dad fell asleep long ago, like at 9:00 at night, and I woke him up and got him in bed, but I- OOUF, damn it I fell outta my chair, well see you guys later, I'm going to move somewhere and collapse.

Good luck on that programming, I read practically every post on the boards, so I think I'm up to date about what's going on with GAE, also, I submitted a ticket on sourceforge, I got over my Submitiophobia(not a word, so don't use it!).

..................Flop...zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzDamn it Swordman get outta my Dream!!!zzzzzzzzzz

[ultifd]

Kinda off topic eh? 
Sleep is important...
Usually the earlier is better...
---
Anyways I didn't even know GAE was still in development...
(well till a few days ago.)
What's being improved or added on?

Thanks. 

[silnarm]

What's being improved or added on?

http://sourceforge.net/apps/trac/glestae/report/6

[softcoder]

Good morning everyone. I haven't given up

I had already decided to go back to the 3.2.2._network branch as merging changes into 0.2.x is a hideous endeavour (the code is VERY different) and was proving to be way more risky.

Please let me know when you have fixes checked into svn and we'll test them out. Perhaps they were related to some of the instability we have been seeing here?

Thanks

[silnarm]

Update:

The cancel command crash is fixed, that required more work poking around than I expected, but it turns out that was most definitely the easy part.

The Sync issue was not directly related to auto-commands. There were in fact several problems, my disabling of the fog of war was the first thing to cause grief, but looking on the bright side, we can now have the option of playing multiplayer games without it... I'm sure people have tried the obvious fix for this before, as I did, with unpleasant results... anyway, that issue has been fixed.

Hailstone then identified a problem with Cpu-Ultras, and that has now been rectified.  The final hurdle was some code it took a while to find, causing the stop command to skip updates at irregular intervals... finally located and removed only moments ago.

So after a marathon "add logging code, test, add more logging code, test, fix, repeat" cycle, I finally got two identical logs... the game is running in sync  .

What a 'holiday' ... at least I have another day and then a weekend

It's a horrendous mess of logging at the moment, and I need to rest my eyes for a bit now... but I'll clean it up and get it committed within, say, 2-3 hours...

Cheers.

Edit: Changes are in, there is still a problem with the minimap on clients (the units are drawn, but it is otherwise completely transparent) that I've yet to find the cause of.
@softcoder: I tried to keep the changes contained, but this wasn't really successful, if you have local changes to your working copy, you'll probably get a conflict or two.

[softcoder]

Ok found a # of problems (and closing in on them shortly). There seems to be a bug related to "command types" and passing the wrong command type id or unit id, also cancelling a command and auto-repair end up crashing the gae. I'll post more detailed info once I have better details.

[silnarm]

Ok found a # of problems (and closing in on them shortly). There seems to be a bug related to "command types" and passing the wrong command type id or unit id, also cancelling a command and auto-repair end up crashing the gae. I'll post more detailed info once I have better details.

Excuse my possibly wishful thinking, but are you sure you updated, and that it was successful? These are exactly the problems I spent all of yesterday resolving!

Auto-repair should be disabled, cancelling a command should not crash, and the command/unit Id mismatches are the sync issues, all sorted out in r478.