Re: [v3] -Weffc++ fixups

[Dhruv Matani <dhruvbird at gmx dot net>]

On Wed, 2004-04-14 at 10:43, Benjamin Kosnik wrote:
> This is a preliminary pass to allow -Weffc++ to be used again. I still
> have to run through the testsuite.
> 
> There are some things that I think are pretty interesting, and a lot of
> additional work to get this to work fully.
> http://gcc.gnu.org/onlinedocs/gcc/C---Dialect-Options.html#C++%20Dialect%20Options
> 
> In general, it would be nice if -Weffc++
> 
> 1) didn't complain when assignment operators return a function that
> returns *this
> 
> 2) if a base class has private copy constructors and assignment
> operators, then don't warn that they are missing in derived classes.
> 
> It would be cool, in some imaginary world, if -Weffc++ could be made
> conditional on a specific item, instead of lumping them all together.
> 
> Still, useful things were found, some of which are listed below. If it's
> possible, I would like to make -Weffc++ play nice with g++ so that
> people can use this feature.
> 
> 1) bvector was not fixed up w/ the allocator patches
> 
> 2) the base allocator classes don't have virtual destructors,
> intentionally, which I think is ok since the destructors don't really do
> anything. If people disagree, this would be a wise time to step up.
> 
> 3) there are a lot of places where member initialization lists should be
> used for libstdc++.
> 
> 4) cxxabi.h was reformatted. I couldn't take it anymore, but this will
> be split out from this patch.
> 
> I'll wait for comments, and don't expect parts of this patch to go in at all.

Just a few questions:

1. Are the virtual dtors meant to make the -Weffc++ pass? I suspect yes.

2. Is there any way to disable parts of code from being checked by
-Weffc++ something like:

#SOME_PREPROCESSOR_COMMAND -Weffc++
//Code that should be excluded from the check.
#END

3. Will the virtual dtors increase the class size under normal
circumstances?

4. bitmap_allocator is giving problems with linking when boostrapped as
the default. Paolo is looking into it, but after seeing your patches, I
have a feeling that it's happening because the explicit instantiations
for the static is not there. Is my hunch valid?


-- 
        -Dhruv Matani.
http://www.geocities.com/dhruvbird/

Proud to be a Vegetarian.
http://www.vegetarianstarterkit.com/
http://www.vegkids.com/vegkids/index.html