[M3devel] compiler build problem

Rodney M. Bates rodney_bates at lcwb.coop
Wed Nov 20 03:13:07 CET 2013


I don't know for sure what is going on in your case, but I have several times found this
error disappearing after exhaustive cm3 -realclean or do-cm3-*.sh realclean, followed
by recompiling everything from scratch.  I think I recall one time even vetting this and
finding some package that was present in different versions in the link closure, or something
similar.  The declaration-granularity automatic recompilation in the compiler only pretends
to work within a package.  Inter-package, you are on your own to make sure everything is
compiled in the right order and consistently.

Sorry, I don't have any help on what order that is.  But I think the order of packages
in pkginfo.txt is, at least mostly, right, if you use the do-cm3-*.sh scripts.

On 11/19/2013 05:47 PM, Coleburn, Randy wrote:
> I’m running into a new problem.
>
> I’ve updated from the HEAD branch and rebuilding everything on Windows XP again.
>
> Building of the package group “front” in stage 1 goes well and I install the new cm3.exe.
>
> Then, when I try to build in stage 2 using the new cm3.exe, I get the following error when invoking cm3.exe:
>
> ***
>
> *** runtime error:
>
> ***    A compile-time type is missing.
>
> ***    file "..\src\Utils.m3"
>
> ***
>
> Stack trace:
>
>     FP         PC      Procedure
>
> ---------  ---------  -------------------------------
>
> 0x12fea0   0x6d6c76  Fail + 0x79 in ..\src\runtime\common\RTType.m3
>
> 0x12fecc   0x6d5ea3  ResolveTypeLink + 0x50 in ..\src\runtime\common\RTType.m3
>
> 0x12fef8   0x6c242b  ResolveTypeLinks + 0x57 in ..\src\runtime\common\RTLinker.m3
>
> 0x12ff2c   0x6c21ab  FixTypes + 0x107 in ..\src\runtime\common\RTLinker.m3
>
> 0x12ff40   0x6c1c9d  AddUnitI + 0xe2 in ..\src\runtime\common\RTLinker.m3
>
> 0x12ff60   0x6c1d56  AddUnit + 0x9c in ..\src\runtime\common\RTLinker.m3
>
> 0x12ff7c   0x401038  <???>
>
> 0x12ffc0   0x70d738  RTTypeMap_M3 + 0x4d6 in ..\src\runtime\common\RTTypeMap.m3
>
> .........  .........  ... more frames ...
>
> Any ideas on what is wrong?
>
> Thanks,
>
> Randy Coleburn
>




More information about the M3devel mailing list