[llvm-commits] [llvm] r169218 - in /llvm/trunk: include/llvm/ lib/CodeGen/AsmPrinter/ lib/VMCore/ test/CodeGen/ARM/ test/CodeGen/Thumb/ test/CodeGen/X86/ test/DebugInfo/ test/DebugInfo/X86/ test/JitListener/

Duncan Sands baldrick at free.fr
Wed Dec 5 09:06:09 PST 2012


Hi Eric,

On 05/12/12 18:02, Eric Christopher wrote:
>
>
>
>
>     is this about a default, used when the language front-end forgot to provide a
>     lower bound?  For Ada there isn't really any sensible default, but 1 is better
>     than 0.  The Ada front-end should just specify explicitly what the lower bound
>     is for each object/type.  For Fortran 1 makes sense, but again better if the
>     front-end is explicit about it.  I don't have time to read the rest of the
>     thread right now, so I'm a bit lost as to what this is all about and why it is
>     related to DIBuilder.  I've been trying hard not to have to understand how debug
>     info works, but it sounds like those halcyon days are drawing to a close!
>
>
> Nah, you're good. The defaults I mentioned above are just fine for the languages,
> I just didn't have a feel for if/how you could have a lower bound of zero for
> fortran/ada array types while the default is 1.

in Ada you can have a zero lower bound, or a negative lower bound.

> The DIBuilder bit was just a quick poke as DragonEgg is the only remaining supported
> user of a lot of code in the debug support. I can probably just change it myself at
> some point as well. :)

:)

Ciao, Duncan.




More information about the llvm-commits mailing list