Iain Buclaw

> On LDC, e.g., if you specify `-target` it won't let you use `-m64` or `-march` to override it and it translates to `-mtriple` directly, but that is possible on...

> I'm rather talking about `ldmd` and `gdmd`. Is, e.g. `gdmd` wrapping `-target` to something like `-mtriple`? [No](https://clang.llvm.org/docs/CrossCompilation.html#cross-compilation-issues). > In the case of LDC, it does, hence my note. See...

> I meant LDC upstream, sorry. @kinke or I can make a patch to fix this, if this is going forward. Well, wrappers are free to do whatever. So nothing...

Bear in mind that I've been an outsider to the review process for saoc in general, and not knowing just how much of the implementation idea was written down in...

> This is solely intended to be available to things like dub so one can depend on multiple versions of a given module in the same dependency tree. I guess...

> Could you elaborate a bit on what you're trying to suggest? When compiling modules present in a directory named `old`, change all references to them as if they came...

> @ljmf00 @ibuclaw what's the status of this PR? The needless whitespace changes make it hard to vet that the functional changes to the C++ headers are OK.

Should be tested against either gdc or ldc (@kinke) to gauge how much of a nuance that asym field will be without an accessor.

This needs rebasing on top of #13862.

> > This needs rebasing on top of #13862. > > Done. Do you want me to make `sym` an accessor? Probably an inline implementation on `mtype.h` is preferable. Could...