Open Caid11 opened 2 months ago
I managed to reproduce this error. It is tricky to minify and often instead runs into the linked #500 instead (eg. not being able to look up world/gameWorld
). On linux, you can reproduce this using my stub implementation of socket.kk and raylib.kk.
The bug occurs since modmapRead
is passed a module with name nameNil
:
bb(1, 1): internal error: modmapRead: empty modname
CallStack (from HasCallStack):
error, called at src/Compile/Build.hs:218:32 in koka-3.1.2-7SWN83ElxwDEblvTDAD6Nq:Compile.Build
modmapRead, called at src/Compile/Build.hs:347:25 in koka-3.1.2-7SWN83ElxwDEblvTDAD6Nq:Compile.Build
moduleWaitForImports, called at src/Compile/Build.hs:307:19 in koka-3.1.2-7SWN83ElxwDEblvTDAD6Nq:Compile.Build
moduleCodeGen, called at src/Compile/Build.hs:262:18 in koka-3.1.2-7SWN83ElxwDEblvTDAD6Nq:Compile.Build
moduleCompile, called at src/Compile/Build.hs:107:25 in koka-3.1.2-7SWN83ElxwDEblvTDAD6Nq:Compile.Build
I added tracing for all modmap(Try)Put
and modmapRead
operations -- the full log is at https://gist.github.com/anfelor/57e63dc6f2b7933bb40a2753c6b54b71 for a minified version of the code that only imports raylib/raylib, socket/socket, server, ui
.
I found this while trying to reduce my reproducer for #500. It's very similar, and could be related or a duplicate.
When I try to compile my code, I see this error:
Similar to #500, passing
--rebuild
avoids the failure. I found that removing thebleh : duration
member of theserver
struct and updating its constructor calls fixes the issue.Notes:
Failing code is here: https://github.com/Caid11/Functional-Program-Studio/tree/internal-compile-error-map/blockbuilder