| Re: [cdt-dev] Build System redo |
Hi all,
I am trying to write a CMake plugin for CDT (see bug 350206, got stuck but
still working on it). So my perspective is influenced by external build
systems like CMake.
I fully agree with idea of the "portable build model" (I also know it as
"project model"). This is somthing that is implemented very nicely in CMake,
but also qmake ,boost.build and the autotools.
Project model is the file "CMakeLists.txt" (under version control).
(Cross-)Compiler for the target platform is defined in a "toolchain file"
outside of the project model (included with cmake distribution or an own one).
Similar: qmake ->project model in .pro file; compiler settings in mksepcs,
autotools -> project model in Makefile.am, toolchain selected during configure
run via --target=arm-linux-gcc)
These build systems share a property which was not mentioned in this
discussion yet: there is a "configure" step involved in the build. Needed only
once to select the toolchain, find external libraries and include paths, but
not for every build during development.
The indexer cannot know the include paths before the configure step was run.
But it would be nice to be able to navigate through the sources without having
to build them before, especially for large projects.
I would really like to see better support for the "configure" step.
best regards
Martin
Am Donnerstag, 19. März 2015, 16:31:40 schrieb Jesper Eskilson:
> By "portable" I mean having a CDT project which can be checked out and
> built on both Windows and Linux (for example) without having to make any
> changes to the project, just as you can do with Java projects. I
> understand that it is most likely not easy to do, though.
>
> /Jesper
_______________________________________________
cdt-dev mailing list
cdt-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cdt-dev