Re: [PATCH] Conditionalize maybe_suppress_debug_info in gcc/cp/search.c

[Mark Mitchell <mark at codesourcery dot com>]

Carlos O'Donell wrote:

> This patch adds the flag "-femit-class-debug-always" which causes all
> the C++ class debug information to be emitted. Even if the vtable is not
> being emitted.

As background, there are debuggers (such as at least some versions of
Wind River's VisionClick) that cannot handle the unification of class
debugging information.  They are unarguably broken, but it's still a
good thing if GCC works with them.

> The documentation is perhaps the best description of this patch.
> ~~~
> This is indirectly used by the dwarf2 backend to determine wether class
> debug information should be suppressed. Using this flag can increase the
> debug information output size by a factor of 2. It can provide class
> debug information in the file that the class was defined, instead of
> just the location where the class is used. This is usefull for debug
> information consumers that lack the ability, or do so poorly, to unify
> program debug information.
> ~~~

You need a spell-checker: "whether" and "useful".

I think you should put the positive parts first -- that this helps you
use some debugger -- before launching into the unappetizing code bloat.

"Instead of emitting debugging information for a C++ class in only one
object file, emit it in all object files using the class.  This option
should be used only with debuggers that are unable to handle the way GCC
normally emits debugging information for classes because using this
option will increase the size of debugging information by as much as a
factor of two."

Technically, I think the patch is fine, but I do not want to approve it
unilaterally.

-- 
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713