Re: [commit] Changes for architecture type vector in Ada

[Andrew Cagney <cagney at gnu dot org>]

> The following patch is intended to bring the Ada sources in line with the
> C sources in their handling of language-specific types.  (When I checked
> this is, I also revised my last ChangeLog entry to give the right, i.e.,
> official, e-mail address).  

Thanks!  Any complaints / suggestions on the interface (I'm not that 
happy with it).

We[gdb community] only have all the others.

> Andrew, at some point I asked you a question relevant to this patch
> for which I seem to have mislaid any answer you may have given: If at
> some point in a language-specific file, one needs to refer to a
> language-specific type, what is the MC (methodologically correct)
> method for accessing that type?  By name, as by
> language_lookup_primitive_type_by_name (seems a bit wasteful to have
> to do a search), or by some other means?  Thanks.

I remember you asking the question, but when I went back I couldn't 
re-find it :-(

Language lookup_primative_type_by_name struck be as just wrong.  I even 
tried to eliminate it but stumbled into a minefield (something to do 
with how C & C++ are a hybrid, the details elude me).

My hunch is that Ada, and the other languages, should instead have 
per-osabi local data (see reggroups for how to do this without needing 
changes to the architecture vector) that contains the language specific 
primitive types that they really need.

--

With this in place, and having looked through the code, I think we've just:

-- strip out the #ifdef GNAT_GDB code
(I'll send you a better way to maintain local changes off-list)

-- enable Ada

left.

Andrew