[PATCH] Add support for DebugFission to DWARF parser

[Eric Christopher]

Ah Right, I didn't make that connection. If you don't mind submitting
that separately I'd appreciate it.

On Mon, Aug 5, 2013 at 12:49 PM, Alexey Samsonov <samsonov at google.com> wrote:
>
> On Mon, Aug 5, 2013 at 11:04 PM, Eric Christopher <echristo at gmail.com>
> wrote:
>>
>>
>>   Few questions inline, otherwise looks pretty good.
>>
>>
>> ================
>> Comment at: include/llvm/DebugInfo/DIContext.h:127
>> @@ +126,3 @@
>> +  enum DIContextKind {
>> +    DICK_DWARF
>> +  };
>> ----------------
>> Unfortunate naming in english. How about CK_DWARF?
>>
>> ================
>> Comment at: lib/DebugInfo/DWARFContext.h:30
>> @@ -29,3 +29,3 @@
>>  class DWARFContext : public DIContext {
>> -  SmallVector<DWARFCompileUnit, 1> CUs;
>> +  SmallVector<DWARFCompileUnit *, 1> CUs;
>>    OwningPtr<DWARFDebugAbbrev> Abbrev;
>> ----------------
>> Consider breaking this part out first in a separate patch? Also, since
>> there's a single DWO for a single compile unit I'm not sure about the change
>> to pointer as far as relevance to this patch. Can you give a bit more
>> explanation here?
>
>
> Sure, I can submit bits of this patch separately (RTTI for DIContext and
> this change). For simplicity, I make OwningPtr<DWARFCompileUnit> DWOCU a
> member of DWARFCompileUnit class (so that each compile unit may store and
> own .dwo associated with it). But then DWARFCompileUnit is not trivially
> copyable anymore, so we can't store it (or OwningPtr<DWARFCompileUnit>) in
> vector anymore.
>
>>
>>
>>
>> http://llvm-reviews.chandlerc.com/D1164
>
>
>
>
> --
> Alexey Samsonov, MSK