Everything works until it doesn't.

Page: 1 2
Jun 30 2018, 11:04 pm
Okay, I think I've found it. There are some items I think are not quite right (like the bad index error in getrev/New() seems to happen from code that's expecting a log file to have more than 1 line, and I don't have that file at all), but otherwise the problem is a simple one.

Among the compiler changes I'd been making, I changed the default arg init code for procs:
// was this
if(var == null) var = default

// now is this
if(isnull(var)) var = default

It takes up less proc real estate and should be faster--but the wrong instruction was popping the value from the stack, so an internal true/false value wasn't getting set correctly. End result: those vars did not get initialized.
Jun 30 2018, 11:22 pm
And now it appears that's only part of the problem after all. Energy weapon inits aren't quite right so I'm looking at that. The problem is with lists controlled by init procs, so it's no mystery where I have to look for that one at least.
Jul 1 2018, 12:28 am
Okay, I finally solved that problem. The only issues I'm getting now come from having disabled rust_g.

There was a minor issue with init procs during analysis still creating objects but not always deleting them properly, and now they won't create objects at all (except stuff like lists, but basically this means no datums, objs, etc.).
Jul 1 2018, 12:30 am
Lummox JR resolved issue with message:
During analysis of init procs to consolidate them, objects could be created that were sometimes not let go of properly or worse, called the soft Del() proc. Now init procs will no longer create atoms, images, or hard datums during anaylysis.
Jul 1 2018, 12:38 am
I forked off the compilation issue to another post and marked both resolved.

My hope was to get 512.1434 out tonight but it's so late now that's not going to happen, so it'll likely be tomorrow night.
Page: 1 2