Re: [PATCH] Handle DW_AT_encoding on DW_TAG_enumeration_type

[Tom Tromey <tromey@adacore.com>]

Tom> Now, we could change update_enumeration_type_from_children to also
Tom> examine the attribute forms, with the idea that if we see DW_FORM_sdata
Tom> then it is definitely signed.  While that wouldn't help this particular
Tom> test, it's possible to construct one where it does.  In fact I happened
Tom> recently to be investigating this for gnat-llvm, where there's a type:

Tom>  [  10cf]      enumeration_type     abbrev: 21
Tom>                name                 (strp) "tag_t"
Tom>                byte_size            (data1) 1
Tom>                decl_file            (data1) p.adb (1)
Tom>                decl_line            (data1) 20
Tom>                alignment            (udata) 1
Tom> ...
Tom>  [  10de]        enumerator           abbrev: 22
Tom>                  name                 (strp) "object"
Tom>                  const_value          (sdata) 18446744073709551615 (-1)

Tom> Here the enumeration type doesn't specify a sign, but there's a constant
Tom> that is clearly signed.  FWIW this occurs in array_of_variant.exp.

Tom> I'll address this in a separate patch.

I still could do this, but I'm not sure how important it really is.

For gnat-llvm, after digging into this, I think this is a compiler bug
and I'm going to modify the compiler to ensure that enumeration types
always have an appropriately-signed base type.

Tom