Re: [PATCH] Further <resolv.h> cleanup

[Adhemerval Zanella <adhemerval dot zanella at linaro dot org>]

On 11/04/2017 07:33, Florian Weimer wrote:
> I'm not sure if this is okay from an API compatibility point of view.
> 
> _res._flags is supposed to be private to the resolver, so the RES_F_* constants should not be in an installed header file, but we nevertheless included them in <resolv.h>.

In a strict way we should continue to provide it (at least on compat way), however
since it is clearly an wrong exposure from glibc and messing with the flag could
lead to unknown side effects I would say this change seems ok.  Also, trying to get
for the flags leads to no meaningful usage on debian codesarch, so I think we should
be safe 

Aas a side note, I also think the effort to keep to maintain a certain level of
api compatibility for this specific issues is just waste effort without much gain. 

> 
> The other cleanups definitions should be uncontroversial.  The function declarations were discovered by comparing the symbol lists of libc and libresolv with an C AST dump after parsing <resolv.h>.
> 
> Thanks,
> Florian