On Thu, Oct 14, 2010 at 2:32 PM, S Roderick <kiwi [dot] net [..] ...> wrote:
> On Oct 13, 2010, at 04:28 , Peter Soetens wrote:
>
>> On Monday 11 October 2010 12:04:36 Ruben Smits wrote:
>>> On Monday 11 October 2010 11:54:31 Peter Soetens wrote:
>>>> One of the problems newcomers have is to create and compile a component
>>>> from scratch in 5 minutes. An (introduced to) orogen user will succeed in
>>>> this, but I was looking for a solution for people that first want to play
>>>> with the code.
>>>>
>>>> The idea is to have a tool that works like this:
>>>>
>>>> $ orocreate-pkg myrobot
>>>> $ ls myrobot
>>>> CMakeLists.txt
>>>> manifest.xml
>>>> myrobot.cpp
>>>> myrobot-types.hpp
>>>> orocos-myrobot.pc.in
>>>> typekit/
>>>>
>>>> which generates a CMakelists.txt, manifest.xml, .pc file and some other
>>>> dumb template code. It will rely on typegen to generate the transport
>>>> c++/cmake code (in the typekit dir). The main actor here is the
>>>> UseOrocos.cmake file that defines all the utility macros you can use in
>>>> this new package.
>>>>
>>>> All components in the toolchain or outside should rely on these macro's
>>>> too, when they're hand-written (ie outside of orogen).
>>>>
>>>> The UseOrocos.cmake file is installed with the RTT's find macros. I was
>>>> wondering where the orocreate-pkg tool should best live: in a new
>>>> repository under the orocos-toolchain, within RTT or with orogen ?
>>>>
>>>> Thoughts ?
>>>
>>> If it supplies duplicate functionality wrt orogen I would not put it in
>>> orogen. I would put it in OCL, to not pollute RTT, and IMHO it belongs
>>> there.
>>
>> The tool depends mainly on RTT (UseOrocos.cmake and rtt headers) and only on
>> one file of OCL ("ocl/Component.hpp"). Since OCL's Component.hpp also only
>> depends on RTT (it doesn't have a .cpp file, it's header only), we're back to
>> the question if Component.hpp belongs in OCL in the first place.
>>
>> I'll dump it in OCL for now, because of that dependency, but it still itches.
>
> +1, but practicality wins sometimes too ...
>
> Is this a first implementation of the CMake "application macros" that we discussed in Barcelona?

Yes. The biggest preparation was done by Adolfo BTW, which had set a
clean stage to start from.

>
> Is this something that OCL could use to work with RTT, instead of more custom CMake logic?

Yes. All (hand written) components, including ocl's should use these macros.

Peter