[CMake] How to describe lib dependencies when a lib has different possible implementations?

[Christopher Harvey]

If you want to compile two execs, one for each static library you can 
use a two element list and go "for_each" over each library and create 
two execs. Otherwise you can just hardcode it into cmake.

Alexandre Feblot wrote:
>  
> Hi,
>
> in fact, in my current system, the implementation to use is pretty hardcoded in each executable makefile, in the link line. There is no automatic lib dependency management.
>
>  
> Alexandre
>
> -----Message d'origine-----
> De : cmake-bounces at cmake.org De la part de Christopher Harvey
> Envoyé : jeudi 19 mars 2009 17:58
>
> Hi,
> Interesting question. You said that
> "only one of them has to be used in every executable, the implementation 
> to use is determined by the executable."
> These are static libraries, so how can the exec pick one if it isn't 
> even built? (As opposed to loading one dynamically)
> Basically I'm still wondering at what point you want the decision for 
> one of the two libraries to made and how you make that decision.
> How exactly did you do it beforehand, without cmake?
> CMake has IF statements, so I'm pretty sure it'll be no problem just 
> have no idea what you would put in those ifs.
>
> Chris.
>
> Alexandre Feblot wrote:
>   
>> Hi,
>>  
>> I'm facing an issue while trying to port my current build system to 
>> cmake and don't know is there is a way to solve it:
>>  
>> In our project, we build a bunch of *static* libs (about 360) and 
>> several executables on them.
>>  
>> * one library, in the middle of the lib dependency chain, has in fact 
>> 2 different implementations. That is to say, there are 2 different 
>> libs with the exact same interface, but they don't use the same 
>> back-end, have different source code, and don't have the same 
>> dependencies.
>> * only one of them has to be used in every executable, 
>> the implementation to use is determined by the executable.
>>  
>> Dependency graph example: "->" means "depends on"
>>  
>> *exe1-A* -> lib1 -> lib3 -> *lib4-A* -> lib5 -> lib7
>> *exe1-B* -> lib1 -> lib3 -> *lib4-B* -> lib6 -> lib8
>> *exe2-A* -> lib2 -> lib3 -> *lib4-A* -> lib5 -> lib7
>> *exe2-B* -> lib2 -> lib3 -> *lib4-B* -> lib6 -> lib8
>>  
>>  
>> I'd like to keep using the standard cmake mechanism of dependency 
>> description in which every lib only defines on which other libs it 
>> depends, and let cmake compute the final link line.
>>  
>> The question is: how can I express the dependencie of lib3 on 
>> lib4 whereas lib3 has only a single description, and we don't know 
>> when describing it, if we'll have to use lib4-A or lib4-B. (and 
>> anyway, depending on the executable, we'll have to use each of them).
>>  
>> Is there any concept of "abstract" target (lib4 in my example) which 
>> could be "instanciated" in the executable description (as lib4-A or 
>> lib4-B in the example)
>>  
>>  
>> Thanks,
>>  
>> Alexandre
>>  
>>
>> This email was sent to you by Thomson Reuters, the global news and 
>> information company.
>> Any views expressed in this message are those of the individual 
>> sender, except where the sender specifically states them to be the 
>> views of Thomson Reuters.
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>>
>> Please keep messages on-topic and check the CMake FAQ at: http://www.cmake.org/Wiki/CMake_FAQ
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.cmake.org/mailman/listinfo/cmake
>>     
>
> _______________________________________________
> Powered by www.kitware.com
>
> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>
> Please keep messages on-topic and check the CMake FAQ at: http://www.cmake.org/Wiki/CMake_FAQ
>
> Follow this link to subscribe/unsubscribe:
> http://www.cmake.org/mailman/listinfo/cmake
>
>
> This email was sent to you by Thomson Reuters, the global news and information company.
> Any views expressed in this message are those of the individual sender, except where the sender specifically states them to be the views of Thomson Reuters.
>
>
>
>