Fix IForward definition

[Kenzzer]

Fixes #2456

This bug was known for a while (9th of April 2026), but unfortunately the lack of issue on the tracker made me forget how urgent it was.
https://discord.com/channels/335290997317697536/335290997317697536/1491812815860797532

This PR hopefully fixes vtable shift bugs with IForward forever. I have removed the inheritance of ICallable by IForward, if any extension out there relied on the ability to mix IForward(s) with ICallable(s) this is unfortunately permanently broken. However I am highly doubtful this is a feature that was desired by extensions.

An alternative fix to this PR would be to have sourcepawn's ICallable be turned into a new class ICallable2. I don't think I need to argue why this is a bad solution and avoided it.

One more fix would be to find a method so that gcc,clang,msvc doesn't merge the vtable of IForward with ICallable. However even if that is possible, this won't help already compiled extensions. So this fix is a no go as well.

[Kenzzer]

Disclaimer I have not tested the following changes, everything was done in the web editor. We desperately need a test suite for this sort of things.

[dvander]

I'm 80/20 convinced now is the right time to delete a ton of crap from the SP API. It's almost 20 years old and we should never have tied ICallable like this... we will need a new native signature to add anything.On May 12, 2026 2:10 AM, Benoist ***@***.***> wrote:Kenzzer left a comment (alliedmodders/sourcemod#2457) Disclaimer I have not tested the following changes, everything was done in the web editor. We desperately need a test suite for this sort of things. —Reply to this email directly, view it on GitHub, or unsubscribe.Triage notifications on the go with GitHub Mobile for iOS or Android. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[dvander]

As noted on IRCDiscord, the fix has to be upstream in SP since it also pushed the IPluginFunction vtables around.