Hard Light Productions Forums

Hosted Projects - FS2 Required => Blue Planet => Topic started by: Rapsodius on February 15, 2019, 05:36:46 pm

Title: Universal Truth (bp2-23.fs2) crash
Post by: Rapsodius on February 15, 2019, 05:36:46 pm
Hi! I recently downloaded BP: Complete from Knossos, to play WiH once and for all (back in my day, only AoA was finished).
But I got errored out on the last mission, of all things D:

https://fsnebula.org/log/5c674c4ccb0d33509f7569c0 (https://fsnebula.org/log/5c674c4ccb0d33509f7569c0)

Thanks!
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: General Battuta on February 15, 2019, 09:01:01 pm
...huh. Did that SEXP get changed? Does it not work correctly on Mac?

Can you try with an older build?
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: Rapsodius on February 16, 2019, 03:09:10 pm
I don't know as I don't have a Mac. I switched to stable build 3.8.0-3 and it loaded the mission. However I lost my campaign process due to mismatch in player files format (nightlies switched to json)
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: General Battuta on February 16, 2019, 04:29:27 pm
Oh my bad. For some idiot reason I saw 'appdata' and assumed it was a Mac, even though it's clearly not.

Could we get a coder to take a look at this log?
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: General Battuta on February 16, 2019, 04:32:38 pm
Also you can 'recover' your progress by just going to the techroom, selecting mission simulator, pressing ctrl-shift-S and playing from there.
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: niffiwan on February 16, 2019, 06:41:24 pm
I've had enough of a look to confirm this occurs on the latest code (i.e. master) however I'm working on a different issue right now. Could we get this reported on github as an issue (https://github.com/scp-fs2open/fs2open.github.com/issues) so it's not lost? Please include the link to the logfile, and a cmdline to reproduce:

i.e.
https://fsnebula.org/log/5c674c4ccb0d33509f7569c0
-mod blueplanetcomplete-1.0.1,MVPS-3.7.2 -start_mission bp2-23.fs2

thanks!
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: praseodym on February 17, 2019, 05:47:22 am
What about 3.7.2 stable?
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: niffiwan on February 18, 2019, 01:52:42 am
IIRC Knossos requires 3.8.0 or newer to work correctly.
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: Rapsodius on February 27, 2019, 10:38:46 pm
What about 3.7.2 stable?

What about it? When I choose Stable release channel in Knossos it dowloads 3.8.0-3, not 3.7.2, so I guess it must've been updated
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: Goober5000 on February 28, 2019, 01:43:18 am
Thanks for opening a GitHub ticket.  This looks like it is related to the fireball.tbl changes I added recently.  Will take a look.
Title: Re: Universal Truth (bp2-23.fs2) crash
Post by: Goober5000 on March 02, 2019, 04:28:14 pm
Copied from the ticket (https://github.com/scp-fs2open/fs2open.github.com/issues/1995)...

Quote
I tried several times to run Blue Planet Complete in a debug build (within the Visual Studio debugger) and kept getting Out Of Memory errors.

However, from a visual inspection of the log and the mod, this is almost certainly a mod bug that was uncovered by the stricter error checking in the new fireballs.tbl.  Note that there have only ever been six default fireballs in the table, numbered 0 through 5; an index of 6 is out of range.  Here are the definitions from 3.8.0...

Code: [Select]
#define FIREBALL_EXPLOSION_MEDIUM 0 // Used for the 4 little explosions before a ship explodes
#define FIREBALL_WARP 1 // Used for the warp in / warp out effect
#define FIREBALL_KNOSSOS 2 // Used for the KNOSSOS warp in / warp out effect
#define FIREBALL_ASTEROID 3
#define FIREBALL_EXPLOSION_LARGE1 4 // Used for the big explosion when a ship breaks into pieces
#define FIREBALL_EXPLOSION_LARGE2 5 // Used for the big explosion when a ship breaks into pieces

There are no fireballs.tbl or *-fbl.tbm tables in bpc-core.vp, and mv_effects-fbl.tbm just modifies the existing entries without adding any new ones.  Thus there could never have been a fireball with an index of 6, even in 3.8.0.

The only change in behavior here is that in the new code, this error is caught when the mission is parsed rather than silently failing at the time the sexp is triggered.