[petsc-dev] cmakeboot.py is a total disaster needs fixing or removal

Barry Smith bsmith at mcs.anl.gov
Mon May 23 16:54:29 CDT 2011


  Jed,

   You were never suppose to see this email. It is null and void. (at least for now :-).

   Barry


On May 23, 2011, at 4:50 PM, Jed Brown wrote:

> On Mon, May 23, 2011 at 23:30, Barry Smith <bsmith at mcs.anl.gov> wrote: 
>    You move a bunch of stuff that has nothing to do with cmake into cmakeboot.py  for example    self.make          = self.framework.require('config.programs',        None)
> 
> Why does this modify anything in the enclosing scope? I used to run cmakeboot.py as a completely separate process just so I could be sure that reading variables wouldn't pollute anything else. But if it's going to run as part of configure, before the RDict is written to a file, then it needs a way to read.
>  
> now if your silly cmakeboot stuff doesn't work then variables like OMAKE are never defined in the generated petscvariables file and so nothing can get built.
> 
> How can cmakeboot.py read variables without modifying _anything_?




More information about the petsc-dev mailing list