[fixed] 3.7.1: Trying to move unit into occupied cell

[tomreyn]

Sadly I have no idea how to reproduce this, but since I never ran into this message bfore, I thought I'd post it nevertheless:

Code: [Select]

megaglest v3.7.1
Compiled using: GNUC: 40403 [64bit] on: Nov 23 2012 00:00:58 platform: Linux-X64 endianness: little
SVN: [Rev: 3948] - using STREFLOP [SSE] - [no-denormals]
radeon: Acquired Hyper-Z.
[2013-02-11 21:45:46] *ERROR* In [game.cpp::update Line: 1790] Error [Trying to move unit [100235 - worker] into occupied cell [x [32] y [102]] and field = 0, unit already in cell [100229 - worker]
Stack Trace:
./megaglest:Shared::Platform::megaglest_runtime_error::megaglest_runtime_error(std::string const&, bool)address [0xb18eae] line: 237
./megaglest:Glest::Game::Map::putUnitCellsPrivate(Glest::Game::Unit*, Shared::Graphics::Vec2<int> const&, Glest::Game::UnitType const*, bool)address [0xa5a77c] line: 1336
./megaglest:Glest::Game::Map::putUnitCells(Glest::Game::Unit*, Shared::Graphics::Vec2<int> const&, bool)address [0xa5ad7c] line: 1262
./megaglest:Glest::Game::World::moveUnitCells(Glest::Game::Unit*)address [0xa1b914] line: 776
./megaglest:Glest::Game::UnitUpdater::updateUnit(Glest::Game::Unit*)address [0xa43d8c] line: 259
./megaglest:Glest::Game::World::updateAllFactionUnits()address [0xa1c2d6] line: 471
./megaglest:Glest::Game::World::update()address [0xa2a36d] line: 569
./megaglest:Glest::Game::Game::update()address [0x642d1c] line: 1544
./megaglest:Glest::Game::Program::loopWorker()address [0x776493] line: 434
./megaglest:Glest::Game::glestMain(int, char**)address [0x768ff6] line: 4793
./megaglest:Glest::Game::glestMainWrapper(int, char**)address [0x76c703] line: 4973
/lib/x86_64-linux-gnu/libc.so.6:__libc_start_main()address [0x7f59f4cae76d] line: 0
./megaglest() [0x5db3e9]address [0x5db3e9]
]
radeon: Released Hyper-Z.

I received this error message when I was playing a game hosted on one of Treba's engineer headless servers on the Five On Three map, playing with three other humans (supposedly all using 3.7.1 on various platforms, located in europe and south america), against three mega CPUs. All other players also received an error message (I was not able to get feedback on which message they received exactly "these messages are too long so I don't read them") at the same time I received mine, so the server seems to have failed.

[titi]

I think thats something I added when I implemented space reservation for morphing units. The code I saw there looked a bit suspicious to me, so I decided to add this errormessage, in case this happens.
Up to now I never saw this error.

We have the following options:
1. we ignore it, as its very rare
2. instead of an error message we fix the problem in the place we now throw an error message.
3. we try to find out whats happened.

3. is very hard to do in my opinion as we cannot reproduce it. 1. gives us the chance to find this bug one day if someone shows something reproducable to us.
2. will fix this exact problem, but maybe hides another problem. BUt its also possible that all this is not a bug , just something that does not happen very often, so 2. would be the right way to handle it.

I think this has something to do with games going out of sync. I have the feeling if we fix all out of sync problems this will be fixed too.

[softcoder]

This could be related to the memory issue we already fixed last week. I would not waste my time on this one unless it can be reproduced on svn head.

[titi]

...I would not waste my time on this one unless it can be reproduced on svn head

me too!