Re: [jamboost] runtime-link

From: David Abrahams (dave_at_[hidden])
Date: 2002-12-05 10:44:24

Vladimir Prus <ghost_at_[hidden]> writes:

> David Abrahams wrote:
>
>>>> <link>shared -- The default (#1 above).
>>>> <link>static -- Internal libraries linked static (#2 above).
>>>> <link>maximum-static -- As many libraries as possible are linked static
>>>>(#3 above)... In some platform, like MacOSX, some libraries must be dynamic
>>>>linked.
>>>>
>>>>Having it this way eliminates the need to worry about what the "runtime"
>>>>libraries are.
>>>
>>>That's an excellent idea! And even if somebody desires to find-tunue
>>>C runtime only, we've a way for extension: just add new values to the
>>><link> feature. Maybe "maximum-static" can be "all-static"?
>>>
>>>I'm eager to implement it as soon as possible. Anybody has objections?
>> 
>> 
>> I think there was much confusion about the meaning of runtime-link in
>> v1; many people thought they should set it to dynamic in order to
>> build a shared library. I am concerned about the name "link" being
>> even more confusable in that same way.
>
> But in case of V2, this is true: you have to set <link> to shared
> in order to build everything as shared libraries. Or set it to
> to static for a specific library to build it as static.

I'm afraid I don't understand what you're saying.

I'm saying this: in v1, people would often do:

lib foo : foo.cpp : <runtime-link>shared ; 

Expecting it to build a shared library. I think you are now saying
that in v2:

lib foo : foo.cpp : <link>shared ;

Does in fact build a shared library. If that's the case, should we not
have a separate feature for specifying how a target should link to
its dependencies?

-- 
David Abrahams
dave_at_[hidden] * http://www.boost-consulting.com
Boost support, enhancements, training, and commercial distribution