Crashes on Specific maps.

[se5a]

regardless of mods.
yes I've tried re-downloading.
this is happening on D-world and throne.

[esses-ss]

Those maps are big (32x32, or greater)?

I can reproduce this issue on 0.82.7.1 (Win) + (BA 7.19 || BA 7.20) + Metal fields map (32x32). In case you cannot reproduce this issue then remove all map related cache files (pathcost etc.) and try again.

0.82.6 works fine.

[abma]

see http://springrts.com/mantis/view.php?id=2295 for updates

[se5a]

Any update on this?
disabling hang detection doesn't make any difference for me.

[se5a]

Is there a work around for this?
I'd really like to be able to play some of these maps again.
(note, for me it's not just really big maps that are having problems, some 28x28 maps also - such as greenheaven)

Edit(greenheaven, not greenfields duh)

[SinbadEV]

Is there a work around for this?
I'd really like to be able to play some of these maps again.
(note, for me it's not just really big maps that are having problems, some 28x28 maps also - such as greenfields)

see http://springrts.com/mantis/view.php?id=2295 for updates

And looking in there it would seem that the issue has been resolved in the most recent revision (but not the officially "released" version).

There was, at one time, a way to test upcoming revisions... the link is here but I'm not sure if the information therein in current:

http://springrts.com/phpbb/viewtopic.php?f=12&t=10555

AKA: Install the latest revision and see if it crashes on big maps for you anymore.

[se5a]

ah cool, thanks for that I'll have to try, I presume this wont work online though?

[SinbadEV]

ah cool, thanks for that I'll have to try, I presume this wont work online though?

Not unless you can find someone to also get the same revision on the test server.

[se5a]

If someone wants to compile the latest rev I'll test it.
it's kind of frustrating that I can't play a large number of the more popular maps.

any idea why this is a problem on some systems but not others?

[abma]

latest rev can be found here:
http://springrts.com/dl/buildbot/defaul ... -g0b381e6/
(or look yourself for the newest rev here: http://springrts.com/dl/buildbot/default/master/ )

[se5a]

Ok,
I copied my spring folder, downloaded
0.82.3-1871-g0b381e6_spring.7z
unpacked it into the copy, tried to run it, got an error, realized I needed some updated Dlls or something.. went looking found DLLs in:
0.82.3-1871-g0b381e6_mingwlibs.7z
(is there some instructions somewhere)
unpacked that into the copy, ran it again. some success this time but:

[f=0000000] Loading Feature Definitions
[f=0000000] Initializing Map Features
[f=0000000] Reading Estimate PathCosts [8]
[f=0000000] Error opening cache\paths\Throne v12339111379.pe.zip
[f=0000000] Spring 0.82+.4.0 (0.82.3-1871-g0b381e6) has crashed.
[f=0000000] Spring 0.82+.4.0 (0.82.3-1871-g0b381e6) has crashed.

infolog.txt

BA 7.20 Throne Fail

(131.62 KiB) Downloaded 108 times

lookeed in cache\paths\ and found that there was no "Throne v12339111379.pe.zip" in that folder.

So it seems to be having problems creating the paths.

So then I tried DSD which is working in the old version.
got further but froze up and I had to force close.

infolog2.txt

BA 7.20 DSD Fail

(90.11 KiB) Downloaded 113 times

tried again just to be sure, and it loaded DSD the second time.

infolog3.txt

BA 7.20 DSD success

(15.13 KiB) Downloaded 117 times

[abma]

uh, sorry. next time please use the installer and install into an empty dir:

for example:
http://springrts.com/dl/buildbot/defaul ... b381e6.exe


can you try to set

HardwareThreadCount=1?

(either edit springsettings.cfg or start spring.exe and select "Edit settings" and then set it there...)

[se5a]

Oh THATs the installer.
right. ok...
trying that and hardware threadcount=1

[se5a]

Wooo!!!

HardwareThreadCount=1

IT WORKS! even in the old (current) version!
Thankyou! I can now play the FFA games!

So what does that do?

[abma]

(some things of these are guesses, because i'm not familiar with this part of the code)
the pathfinder caches data for maps and creates this data with multipele threads. the count of threads is detected as count of cpu cores. the memory usage seems to depend on the mapsize * threads.

as you've 12 cpu-cores... this seems to be only an out of memory.

hardwarecthreadcount overwrites the automaticly detected value.

[abma]

i made a bug report:http://springrts.com/mantis/view.php?id=2436

[se5a]

Ahhh, yes that makes sense.