Re: C++/Java regressions

[Daniel Jacobowitz <drow at mvista dot com>]

On Tue, Nov 25, 2003 at 09:49:50AM -0500, Michael Chastain wrote:
> Good morning,
> 
> I see the same regressions that David does.  This happened in all
> my gcc-3 configurations.  If you want a specific configuration:
> 
>   target = native, host = i686-pc-linux-gnu, osversion = red-hat-8.0
>   gdb = HEAD 2003-11-25 04:21:57 UTC
>   gcc = 3.3.2
>   binutils = 2.14
>   glibc = 2.2.93-5-rh
>   gformat = dwarf-2
>   glevel = 2
> 
> I have two issues.
> 
> === Issue #1: the demangler changed its interface.
> 
> The demangler used to return parameters all the time, but now it looks
> like it returns parameters only if DMGL_PARAMS is given.
> 
> This change did not happen with "complete rewrite".  It happened
> with the followup patch:
> 
>     2003-11-22  Ian Lance Taylor  <ian@wasabisystems.com>
> 
>     * cp-demangle.c (d_encoding): Add top_level parameter.  Change all
>     callers.
>     (print_usage): Display new -p option.
>     (long_options): Add --no-params.
>     (main): Accept and handle -p.
> 
> So the questions are:
> 
> (A) Does gdb want to keep getting the parameters?
>     (My answer: yes we do).
> 
> (B) If we want the parameters, should we change gdb to set DMGL_PARAMS,
>     or should we ask the demangler to keep giving us the parameters with
>     our existing flags, which is usually just DMGL_ANSI?  (My answer:
>     let Ian tell us what the interface actually is, and then we'll adapt
>     gdb to match).

The former, since ths has been the documented interface to the
demangler forever.  I believe the GNU v2 demangler supported it.  But I
thought we did everywhere - where aren't we that this caused a problem?

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer