[boost] Re: Re:Re: [serialization] lack of "os.close()" in many test files- locale_saver

From: Robert Ramey (ramey_at_[hidden])
Date: 2003-12-19 18:12:15

>> This suggests to me that the state_saver library is somehow imcompatible
>> with STL Port in that it doesn'nt "unimbue".  This leaves the locale pointer
>> unchanged in the output stream which crashes in its destructor.
>>
>> In my view this points to difference between STLPort and other implementations
>> of the standard library.

>> I'm not sure of the best place to address this.  Can anyone help here?
>
>> Robert Ramey

>>From STLport forum I recive only next ansver:

>---------------------------------------------------------------
>    Ulrich Eckhardt wrote:

>    This topic is not new. I even think that the standard is unclear on
>this. Relevant discussions took place here (sorry, no link handy), but a
>good place to ask would be on the Usenet in comp.lang.c++.moderated or
>comp.std.c++. There also was a discussion on the boost mailinglist. Search
>the web.

>    However, current behaviour of STLport is to not allow changes to the
>facet after any IO was done. Last time I asked, the consensus was AFAIR that
>this is not the most beautiful thing in the world but still permitted by the
>standard.

Doesn't all this suggest the the locale_saver should be enhanced to 
encapsulate this?

Robert Ramey