Re: [boost] [serialization][tru64cxx]assertionfailedatextendedtypeinfo.cpp

From: Markus Schöpflin (markus.schoepflin_at_[hidden])
Date: 2006-03-16 04:33:03

Markus Schöpflin wrote:

> Joaquín Mª López Muñoz wrote:

[...]

>> I think globally changing to -pt is not worth the potential hassle: after all, it's
>> just one test that's having problems. What I can do is tweak my Jamfile
>> (you're using v1, right?) so as to use -pt in this particular case, only.
>> Would you be available to locally test such a Jamfile?
> 
> I'm using both V1 and V2, but the release will be done with V2 I believe.
> 
> But I'm not sure if changing the instantiation model for just one test is a 
> good choice. Let me try the global switch to -pt first, maybe it will look 
> better this time...

Joaquín,

currently I think the most viable solution is to mark the failure as 
expected. I'm still in the progress of evaluating a switch to -pt, but this 
will take more time.

Maybe you can put an explanation into the markup, something like: "This 
test fails because of limitations in the template instantiation model of 
the compiler used (-timplicit_local), which causes two static objects of 
the type 
extended_type_info_typeid<boost::multi_index::detail::index_node_base<int> 
 > to be created. It could be made to pass by using another template 
instantiation model, but this is currently not feasible."

Feel free to reword it to suit your taste... :-)

Markus