Re: [lmi] Re[2]: enum types rewrite question

[Greg Chicares]

On 2008-08-02 00:01Z, Vadim Zeitlin wrote:
> On Sat, 12 Jul 2008 19:43:23 +0000 Greg Chicares <address@hidden> wrote:
> 
> GC> But I thought I should point out that these files:
> GC>   inputillus.?pp
> GC>   inputs.?pp
> GC>   inputstatus.?pp
> GC>   xenum*.?pp [eight files]
> GC>   xrange*.?pp [five files]
> GC> will be deleted within one month at most.

That's still the plan. It will happen soon.

>  I wonder if this means anything is going to change for the definition of
> mcenum_dbopt type and/or its use in yare_input.cpp or if this code is
> already the new version (as I suspect is the case unfortunately)?

Yes, it's the new version. However, you have proposed a change to
class template mc_enum that I still have not considered yet. I've
glanced at it. I think I've understood the basic idea. It's quite
a broad change to code that's already valid C++, so of course I
hesitate to apply it. First, I would want to give some thought
myself to whether the problem can be solved in a way that does
less violence to the code. But I can't make time for that right
now. Yet I don't want to reject your suggestion lightly, and that
is why I haven't answered the fundamental question yet.

>  I'm asking because, as usual, I have a problem with MSVC compilation
> (sorry for the long lines but I'd rather preserve the error message
> exactly):
> 
> yare_input.cpp(171) : error C2784: 'std::vector<Number> 
> `anonymous-namespace'::convert_vector_type(const 
> std::vector<tn_range<Number,Trammel>> &)' : could not deduce template 
> argument for 'const std::vector<tn_range<_Ty,_Ax>> &' from 'const 
> std::vector<_Ty>'
>         with
>         [
>             _Ty=mce_dbopt
>         ]
>         yare_input.cpp(53) : see declaration of 
> '`anonymous-namespace'::convert_vector_type'
> 
> I'm going to look into fixing this but first I'd like to know if this is
> not something that is going to be changed [soon] anyhow?

No, the part I aim to delete is the part that compiles with msvc.

However, this seems to be the same template-argument-deduction
problem that msvc has with class template mc_enum in general, so
I think it's just another manifestation of the same problem. And
that problem must be attacked at the root. I don't have time for
that now, though.

At this moment, it is probably unnecessary for you to use the
very latest lmi HEAD. I think you and I are working in different
areas of the code, so that your work is probably independent of
mine for a short while. I still have plenty to change, and it'd
probably be a mistake to ask you continually to maintain an msvc
branch that may take a lot of work to keep synchronized with
everything I'm doing right now.