search for: mutuallyincompatible

...rote: > > Slightly formatted/commented patch. > > WDYT? > > This seems to work fine, except that reading a field from a const > AttrConst is not a constant expression in C++03, so the "set" > declarations (ParameterOnly, FunctionOnly, VarArgsIncompatible, > and MutuallyIncompatible) still require a global constructor. > OMG, yes, indeed. > You > can split the values into separate 'const uint64_t' declarations and > use those; it makes the header even uglier, but it works. > > like this (see the new patch)? Yes, that's great, thanks! John....

..., at 2:07 PM, Kostya Serebryany wrote: > Slightly formatted/commented patch. > WDYT? This seems to work fine, except that reading a field from a const AttrConst is not a constant expression in C++03, so the "set" declarations (ParameterOnly, FunctionOnly, VarArgsIncompatible, and MutuallyIncompatible) still require a global constructor. You can split the values into separate 'const uint64_t' declarations and use those; it makes the header even uglier, but it works. John.

...y formatted/commented patch. >> > WDYT? >> >> This seems to work fine, except that reading a field from a const >> AttrConst is not a constant expression in C++03, so the "set" >> declarations (ParameterOnly, FunctionOnly, VarArgsIncompatible, >> and MutuallyIncompatible) still require a global constructor. > > OMG, yes, indeed. > >> You >> can split the values into separate 'const uint64_t' declarations and >> use those; it makes the header even uglier, but it works. >> > > like this (see the new patch)? > > &...

...wrote: > > Slightly formatted/commented patch. > > WDYT? > > This seems to work fine, except that reading a field from a const > AttrConst is not a constant expression in C++03, so the "set" > declarations (ParameterOnly, FunctionOnly, VarArgsIncompatible, > and MutuallyIncompatible) still require a global constructor. OMG, yes, indeed. > You > can split the values into separate 'const uint64_t' declarations and > use those; it makes the header even uglier, but it works. > like this (see the new patch)? --kcc > > John. > -------------- n...

...formatted/commented patch. >> > WDYT? >> >> This seems to work fine, except that reading a field from a const >> AttrConst is not a constant expression in C++03, so the "set" >> declarations (ParameterOnly, FunctionOnly, VarArgsIncompatible, >> and MutuallyIncompatible) still require a global constructor. >> OMG, yes, indeed. >> You >> can split the values into separate 'const uint64_t' declarations and >> use those; it makes the header even uglier, but it works. >> >> like this (see the new patch)? > > Yes...

Slightly formatted/commented patch. WDYT? --kcc On Tue, Feb 7, 2012 at 1:29 PM, Kostya Serebryany <kcc at google.com> wrote: > > > On Tue, Feb 7, 2012 at 12:53 PM, Kostya Serebryany <kcc at google.com> wrote: > >> I see the problem. >> Let me try to come up with a solution that does not involve constructors >> but also does not sacrifice type safety.

On Tue, Feb 7, 2012 at 12:53 PM, Kostya Serebryany <kcc at google.com> wrote: > I see the problem. > Let me try to come up with a solution that does not involve constructors > but also does not sacrifice type safety. > > > I have a patch that uses a proxy POD type. 'make && make check' passes. It's a bit ugly in the header file