Re: [boost] [Boost-users] [typeindex v3.0] Peer review begins Mon 21st ends Wed 30th

Subject: Re: [boost] [Boost-users] [typeindex v3.0] Peer review begins Mon 21st ends Wed 30th
From: Andrey Semashev (andrey.semashev_at_[hidden])
Date: 2014-04-28 06:18:15

On Monday 28 April 2014 06:03:16 Rob Stewart wrote:
> On April 28, 2014 5:54:41 AM EDT, Andrey Semashev 
<andrey.semashev_at_[hidden]> wrote:
> >On Monday 28 April 2014 05:48:42 Rob Stewart wrote:
> >> On April 27, 2014 1:47:22 PM EDT, Andrey Semashev
> >
> ><andrey.semashev_at_[hidden]> wrote:
> >> >On Sunday 27 April 2014 13:02:41 Rob Stewart wrote:
> >> >> On April 27, 2014 10:43:27 AM EDT, Andrey Semashev
> >> >
> >> ><andrey.semashev_at_[hidden]> wrote:
> >> >> >On Friday 25 April 2014 09:25:45 Dominique Devienne wrote:
> >> >> >> But what bothered me a little was the note about the fact that
> >> >> >> pretty_name() was not consistent across platforms/compilers.
> >> >> >
> >> >> >There is generally no way to obtain a portable type name. Each
> >> >
> >> >compiler
> >> >
> >> >> >and
> >> >> >platform has its own mangling rules and there are no rules for
> >
> >the
> >
> >> >> >demangled
> >> >> >type names whatsoever.
> >> >> 
> >> >> The language standard specifies rules for the demangled names,
> >
> >does
> >
> >> >it not?
> >> >
> >> >No. The type_info::name() method returns some implementation-defined
> >> >string
> >> >with no guarantees on its format. __func__ also has unspecified
> >
> >format.
> >
> >> >To my
> >> >knowledge, there are no other sources of type name strings in the
> >> >language.
> >> 
> >> I wasn't clear. I meant that the language grammar specifies the
> >
> >syntax of
> >
> >> the demangled names, though it permits lots of whitespace variation.
> >
> >The problem is that those strings are not required to follow that
> >syntax. And
> >they don't in some cases.
> 
> If they have the required information, however it may be encoded, all that's
> needed is to resemble it. Is that out of scope?

As I said, the format of the strings is not defined. You may be able to 
reconstruct something resembling a C++ type declaration or you may not. In 
some cases there simply isn't a C++ syntax to represent a type name (e.g. 
types with anonymous namespaces). The only way to deal with it is to support 
every particular compiler out there and hope it provides enough information in 
the strings. A task I called not feasible and not particularly useful.