#1000 Evaluate build system migration from autotools to CMake
|Reported by||Christoph Grüninger (firstname.lastname@example.org)|
|Reported at||Dec 20, 2011 16:08|
|Version||Git (pre2.4) [autotools]|
|Operating System||Unspecified / All|
|Last edited by||Christoph Grüninger (email@example.com)|
|Last edited at||Sep 28, 2014 10:35|
|Closed by||Christoph Grüninger (firstname.lastname@example.org)|
|Closed at||Sep 28, 2014 10:35|
|Closed in version||Unknown|
|Comment||Evaluation is over, time for fine-tuning and heavy testing.|
It's December, thus it is time to our yearly CMake discussion. But instead of only discussing, Andreas L. and Christoph G. are willing to make the effort to really migrate the build system to CMake.
The advantages are clear
- provides faster builds
- easier to write and debug (Personally, I dislike working with the autotools)
- new users are less scared compared to using autotools
The disadvantages are clear as well
- It is work, has to be done.
- Absorbs attention from other topics.
- Will introduce new bugs.
- autotools works so far.
But still we are willing to start the project. It won't be done this year but more like next year - or it might even take several Dune versions before we can replace autotools completely.
- What has to be done to get some CMake files included in the official trunk (as experimental CMake support)?
- What CMake version we should rely on?
- Is CMake available on super computers like JuGene or Hermit?
- Can we share tests between autotools and CMake?
- Who is willing to support us?
An overview, the past discussions and some resources were compiled at http://users.dune-project.org/projects/cmakedune/wiki/Wiki