Re: some gcc warnings

> > Therefore I would propose to go with the union approach: > > Yes, that sounds like the best way to fix the problem now. Thanks for > explaining it. OK, I've pushed it. Bruno

update C++ signature tests

Many of the C++ signature tests have not been updated regularly since a long time. I'm adding the missing declarations and also 4 new modules: inttypes-c++-tests strings-c++-tests sys_resource-c++-tests utime-h-c++-tests Bruno >From 89b335be47327cbfb9e6da6b0cc0c2c07bdf7e4c Mon Sep 17 00:0

Re: [PATCH] C++: "#define timeval rpl_timeval" -> typedef in GNULIB_NAMESPACE

Pedro Alves wrote in : > With GNULIB_NAMESPACE, we can do better: Make gnulib define > GNULIB_NAMESPACE::timeval as a typedef to whatever "struct timeval" is > (which is either the system's timeval or gnulib's rpl_timeval > repl

Re: some gcc warnings

On 05/05/2017 09:30 AM, Bruno Haible wrote: Therefore I would propose to go with the union approach: Yes, that sounds like the best way to fix the problem now. Thanks for explaining it.

Re: some gcc warnings

By the way, the proposed patch not only fixes the gcc warnings about aliasing, but also the alignment issue: a declaration of a local variable char block[16]; does not ensure that 'block' will has the necessary alignment for uint32_t accesses. Bruno

Re: some gcc warnings

Hi Paul, > On 05/04/2017 03:24 PM, Bruno Haible wrote: > > https://dev.gnupg.org/rCf17d50bbd31b1faa24af1e46c10bba845becf585 > > https://dev.gnupg.org/rCdfb4673da8ee52d95e0a62c9f49ca8599943f22e > > > > But this fix is only effective for GCC. How could a compiler-independent fix > > look like? > >