Closed John-Colvin closed 4 years ago
when i went in ran maually:
C:\Users\jcolvin_sym\Git\dlang\work\dl\dmd-2.079.0\dmd2/windows/bin\dmd.exe -m64 -of..\generated\build.exe -debug build.d
(note -m64
) and then
..\generated\build.exe --called-from-make "OS=windows" "BUILD=release" "MODEL=32" "HOST_DMD=" "HOST_DC=C:\Users\jcolvin_sym\Git\dlang\work\dl\dmd-2.079.0\dmd2/windows/bin\dmd.exe" "DDEBUG=" "OBJ_MSVC=" ..\generated\windows\release\32\dmd.exe
a digger.exe rebuild
succeeded.
This looks like a bug in DMD's build scritps (build.d
or makefiles). Do you think you could check and see if the problem occurs when you follow the steps Digger does manually?
Pending fixes in DMD to address this issue: https://github.com/dlang/dmd/pull/10490 https://github.com/dlang/dmd/pull/10491
This problem makes digger bisect quite unusable.
Any advice for a workaround?
Any advice for a workaround?
Maybe use another OS? Digger's backend is used for dtest.dlang.io on Linux/64, so we know that should work for all points in time when master and stable aren't visibly broken.
Any advice for a workaround?
Maybe use another OS? Digger's backend is used for dtest.dlang.io on Linux/64, so we know that should work for all points in time when master and stable aren't visibly broken.
I always do when I can, but in this case I'm attempting to narrow down windows 32bit specific regressions
Should be fixed.
@John-Colvin @Robert-M-Muench Is it working for you?
Sorry, a bit late... building the latest Digger works and using it to build the latest DMD works too.
Started with a clean folder