[petsc-dev] [petsc-maint #44162] Why do I always have to run 'conftest' on a batch system now?

[Satish Balay]

pushed the reconfigure.py change to petsc-3.1 [and dev] - along with a
portability fix to reconfigure_PETSC_ARCH.py.

I'll leave the configure.log as-is as its also there for easy
copy/paste for shell invocation of options..

Satish

On Fri, 2 Apr 2010, Barry Smith wrote:

> 
>  Satish,
> 
>    You changed the output of $PETSC_ARCH/conf/reconfigure*.py to print one
> option for line, very readable!
> 
>    Could you please change the generation of reconfigure.py from conftest (for
> the batch option) and the printing of initial configure options in
> configure.log to do the one option for line please? They are too hard to read
> for lots of options.
> 
>    Thanks
> 
>    Barry
> 
> 
> On Apr 2, 2010, at 4:51 PM, Satish Balay wrote:
> 
> > --known-sizeof-MPI-Comm=4 --known-sizeof-MPI-Fint=4
> > 
> > You have typos here. Since some options are missing - configure will
> > generate a conftest file.
> > 
> > Note: think of the generated reconfigure.py or
> > reconfigure_PETSC_ARCH.py as something you can use as your initial
> > script. [with a bit of edits to pretty up and remove absolute paths]
> > 
> > Satish
> > 
> > On Fri, 2 Apr 2010, Richard Tran Mills wrote:
> > 
> > > Hi Folks,
> > > 
> > > It seems that for some time (I know Glenn complained about this a while
> > > ago),
> > > I have been having to run the 'conftest' file that configure.py generates
> > > for
> > > me on Jaguar to get a reconfigure.py that lets me do the actual
> > > configuration.
> > > This is despite the fact that I believe I have everything that conftest
> > > finds out already specified in my configure.py command.  It used to be the
> > > case that I could just make sure all of this stuff got passed to
> > > configure,
> > > and it would not generate the conftest executable for me.  Can someone
> > > tell me
> > > what I am missing?
> > > 
> > > configure.log and the generated reconfigure.py are attached.
> > > 
> > > Thanks,
> > > Richard
> > > 
> > > 
> > 
> > 
>