Closed russel closed 6 years ago
The linker messages you get strangely look like those i have in an unrelated D project. I suspect a compiler or maybe phobos bug, it cant be randomly the same symbols: mine are
dictionnary_suffixarray.d:(.text._D3std4conv__T8textImplTAyaTQeTQhTQkTQnTkZQBcFNaNbNfQBcQBfQBiQBlkZQBq[_D3std4conv__T8textImplTAyaTQeTQhTQkTQnTkZQBcFNaNbNfQBcQBfQBiQBlkZQBq]+0x151) : référence indéfinie vers « _D3std4conv__T7toCharsVii10TaVEQBd5ascii10LetterCasei1TkZQBrFNaNbNiNfkZSQCsQCr__TQCpVii10TaVQCki1TkZQDiFNaNbNiNfkZ6Result »
dictionnary_suffixarray.d:(.text._D3std4conv__T8textImplTAyaTQeTQhTQkTQnTkZQBcFNaNbNfQBcQBfQBiQBlkZQBq[_D3std4conv__T8textImplTAyaTQeTQhTQkTQnTkZQBcFNaNbNfQBcQBfQBiQBlkZQBq]+0x16d) : référence indéfinie vers « _D3std4conv__T7toCharsVii10TaVEQBd5ascii10LetterCasei1TkZQBrFNaNbNiNfkZ6Result5emptyMFNaNbNdNiNfZb »
dictionnary_suffixarray.d:(.text._D3std4conv__T8textImplTAyaTQeTQhTQkTQnTkZQBcFNaNbNfQBcQBfQBiQBlkZQBq[_D3std4conv__T8textImplTAyaTQeTQhTQkTQnTkZQBcFNaNbNfQBcQBfQBiQBlkZQBq]+0x184) : référence indéfinie vers « _D3std4conv__T7toCharsVii10TaVEQBd5ascii10LetterCasei1TkZQBrFNaNbNiNfkZ6Result5frontMFNaNbNdNiNfZa »
dictionnary_suffixarray.d:(.text._D3std4conv__T8textImplTAyaTQeTQhTQkTQnTkZQBcFNaNbNfQBcQBfQBiQBlkZQBq[_D3std4conv__T8textImplTAyaTQeTQhTQkTQnTkZQBcFNaNbNfQBcQBfQBiQBlkZQBq]+0x1c5) : référence indéfinie vers « _D3std4conv__T7toCharsVii10TaVEQBd5ascii10LetterCasei1TkZQBrFNaNbNiNfkZ6Result8popFrontMFNaNbNiNfZv »
as you can see the template instance is different but it's the same template.
So i'm gonna close this because this is not a D-YAML bug.
If it's the same bug then when using the LDC based on latest 2.074.x front-end you should see the problem disappear.
|> ldc2 -v
binary /usr/bin/ldc2
version 1.8.0 (DMD v2.078.3, LLVM 5.0.2)
config /etc/ldc2.conf (x86_64-pc-linux-gnu)
OVERVIEW: LDC - the LLVM D compiler
Can you try with this one https://github.com/ldc-developers/ldc/releases/tag/v1.4.0 ?
Probably not as the Debian package is at 1.8.0.
Maybe someone will have a workaround after all, but it's certainly the same issue.
I'll raise it on the LDC list, to see if there is some insight to be had there.
I close. Now it's sure that it's a compiler front-end issue. With DMD the "-allinst" option works as workaround, if LDC has an equivalent option try it.
Anyone know how to solve this one?