[0.10.0] new save failed after crash

[FreeER]

so I've just encountered this problem while 'testing' my mods (to be honest I got carried away playing with poison ammo and mind control and how great it feels to wipe out biter bases...) anyways, fresh game in 0.10.0 on win 7 x64, busy killing some biters and Factorio crashed with 'not responding'. Reopen and _autosave1 'failed', loaded up _autosave3 and immediately saved it as _autosave3.1, went back to killing (hoping to duplicate it and maybe see anything specific to cause it) and it happened again, but no idea as to why So here's _autosave1, _autosave3 (loadable), and _autosave3.1 (you'll notice that 1 and 3.1 are only ~3 mb compared to 3 which is ~9-10). oh...um there's also a 'clone.zip' save from much earlier that I found interesting because some mind controlled biters were moving back and forth between the two testing 'bases' I'd setup (plus some were 'randomly' dieing for some unknown reason)... and the mods I was using as well if that helps
so... saves+mods zip: ~25 mb in total
Hopefully those can help unearth a bug or two

[Drizzt321]

I encountered something similar at one time. I had a crash during an auto-save, and when I restarted the auto-save had been corrupted and couldn't be read.

[kovarex]

Couldn't it be related to the bad-allocation bug? Are you using 32bit system? Are these saves rather large?

[FreeER]

Couldn't it be related to the bad-allocation bug? Are you using 32bit system? Are these saves rather large?

I'm using 64 bit and the largest save of that world (the one that wasn't corrupted) was only about 10 mb (brand new game), so not particularly large no. Bad-allocation seems doubtful on a 64bit machine with 6 GB of ram (+6GB Swap), and no specific error when crashing (just "Factorio has stopped responding"), at least when you mentioned it being related the address space of 32bit systems and saves being loaded in one block....possible maybe, but doubtful in my unprofessional opinion

[kovarex]

I'm moving it to pending, until it can be reproduced in 0.10.3+ version, I believe it had something to do with the large memory blocks anyway, as it was the change made to 0.10.0 and fixed later.