Re: [jamboost] Re: every static library built twice ?

From: Rene Rivera (grafik.list_at_[hidden])
Date: 2003-12-20 17:27:23

David Abrahams wrote:

> "John Maddock" <john_at_[hidden]> writes:
> 
> 
>>>>using the vc7.1 toolset I get files like:
>>>>boost_date_time-vc71-mt-s-1_31.dll
>>>>boost_date_time-vc71-s-1_31.dll
>>>>boost_thread-vc71-mt-s-1_31.dll
>>>>I'm having trouble wrapping my brain around the concept of a static dll
>>
>>or
>>
>>>>what it would mean
>>>
>>>Dll's shouldn't be built with a static runtime - it generally leads
>>>to all kinds of memory errors - there should be a
>>><runtime-link>dynamic requirement on all the dll targets in the
>>>Jamfiles, if there aren't it's a bug IMO.
> 
> 
> I don't think it would be a bug for Boost.Python. Any DLL that
> doesn't pass resources outside its boundaries can be built with a
> static runtime IIUC.

And if you happen to be building DLLs with a different runtime than the 
executable it would also make sense. Which would happen if you do a 
Boost.Python extension with CW8 but it runs within a Python which is not built 
with CW8.

>>This issue should be fixed in cvs now (I've also re-enabled
>>install-stage for date_time lib on NT, as this does actually work
>>now).
> 
> 
> This seems like a slightly drastic step to take without some
> discussion first.

And the re-enabling of the DLL for date_time doesn't work for CodeWarrior.

-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera_at_[hidden] - grafik_at_[hidden] - 102708583_at_icq