[cmake-developers] CMake master slowdown in generation step

[Nils Gladitz]

On 04.04.2014 17:52, Nils Gladitz wrote:
>
> 8cd113ad1d715cc9ce865956870cd462d3659089
>   Makefile: 0m20.261s
>   Ninja:    2m26.905s
>
> 042c1c834e69ee60b605b02bad8be87b2193a7d2
>   Makefile: 0m21.346s
>   Ninja:    2m29.916s
>
> stage/optimize-source-file-processing
>   Makefile: 0m11.575s
>   Ninja:    0m11.002s
>
> The slowdowns in the two commits seem to be unrelated in the way thy 
> are triggered and the test case only shows the slow down in the first 
> commit.
> The Makefile generator slow down is less prominent than Ninja but it 
> is still present for both.
>

I've timed the slowdown for the second commit (and its surrounding 
commits) within my project:

v3.0.0-rc3:
     Makefile: 0m4.477s
     Ninja:    0m3.899s

aa0a3562dd47bdd6d9ca3058bd1dfd525e79d36d
     Makefile: 0m15.074s
     Ninja:    0m23.151s

042c1c834e69ee60b605b02bad8be87b2193a7d2
     Makefile: 0m15.049s
     Ninja:    0m22.990s

fdcefe3c42fefdfcd27e6cf7b679f39cb99db4b9
     Makefile: 0m10.829s
     Ninja:    0m18.853s

I am not sure how to construct a test case for this one.
     042c1c834e69ee60b605b02bad8be87b2193a7d2 - cmTarget: Compute 
languages from object libraries on demand.

I do have exactly one object library with a single source file in it.

Nils