[Fix inside]Camera + Morph = Crash

[Major Cooke]

Correct. Since it's no longer safe to use tracer for this, a different variable is needed.

Is there a particular property or flag that's set on a morphed actor to safely check, to transition them into a MorphedTarget/Master/Tracer?

[Graf Zahl]

No. There's only a flag on the unmorphed version, but you need tracer to get there. My solution here would be simple:

Add a new variable and set the unmorphed monster both as that and as tracer. I think that will solve all potential issues.

[Major Cooke]

I believe target and master should also do the same thing because I could have been using SXF_ISTARGET as well.

Alright. I'll work on it, if edward isn't.

[Edward-san]

Yeah, you can go ahead.

[Major Cooke]

Bugfix PR. Things turned out to be a tad bit different -- but for good reason.

Tracer is not a safe pointer at all anymore, for storing players and their morphed counterparts. Why? A_RearrangePointers just to name one. The easiest thing to do for players is to introduce the 'alternative' pointer into the mix. I've thus removed all form of storing (un)morphed tracers in the players and morphs involving the morph swappage, and is now done through the alternative pointer.

Monsters remain unchanged: they have an (Un)MorphedMe pointer and aren't of concern.

This fixes several things, and does me a bonus favor:

• Fixes the crash experienced here.
• Fixes a long outstanding bug that only I was able to reproduce a long time ago involving morphs failing to unmorph due to the player tracer pointer being manipulated (Do an A_RadiusGive to all playerpawns which causes them to forget their tracers... That unmorph is never coming!)
• The original target and tracer are now transferred forth for maintained consistency and better tracking of pointers. When the morph expires, it transfers the target and tracer back which may or may not have changed over time, again for consistency.

[Edward-san]

I'm not convinced about the variable name, but dunno about any other 'alternative' ones. I wonder if it would be better to condense all the morph data into a struct or class...

[Major Cooke]

I'm gonna stick with Graf's route and not fix what isn't broken. As for the name...

[Major Cooke]

Thanks for applying this. This can be closed.