1.1 Bug Report: Cross-mod code contamination due to <Globals

[pixelfck]

I found a very nasty bug, maybe it is a known issue, but I could not find any info on it so I'll just post it here.

Bug description:
When there is tLisp code in a <Globals> block, the script is loaded, even if the mod is *not* selected at game start.
Since the mods sorted by UNID, the last mod is the one to take prevailance, the <Globals> block from this last mod will always overwrite the ones from mods before it, whether the mod is selected at game start, or not.

Bug replication code download:
download


How to reproduce:
1. Download the bug replication mod;
2. Clear all mods from the 'Extensions' folder;
3. Install the two mods;
4. Start a new game with 'Test Mod A v1.0' selected and 'Test Mod B v1.0' *un*selected;
5. Select the Debug-class yacht (the modified Sapphire) and start the game;
6. Press 'U' (use) and choose the 'Test ROM':

Expected result:
An on-screen message saying: "Hello, I'm the Test ROM from Test Mod A!"

Actual result:
An on-screen message saying: "Hello, I'm the Test ROM from Test Mod B!"

Cheers

[PM]

EDIT: Nevermind. (Did not fully read everything carefully.)

[george moromisato]

I'll look into it and see how hard it is to fix.

<Globals> were originally considered to be, well, global, so they were not expected to clash. But of course, that's not possible with all the different extensions being created.

[pixelfck]

I understand where the bug comes from. the globas tag is not identified by a UNID, so the engine has no method, other than file inclusion-chain to identify where a global block belongs to.

But I guess you can image the frustration it took to track this one down

~Pixelfck

[george moromisato]

I believe I have fixed this in 1.2 Beta 2. Please try it out.

The fix is to re-run the <Globals> for the selected extensions/adventure at game-start and game-load time.

[pixelfck]

Thanks.
I read so in the changelog, but did not yet have time to check it out. (But I did remove the 'mod' from xelerus.)