[petsc-dev] Is ./configure --help broken?

[Matthew Knepley]

On Fri, Mar 16, 2018 at 9:13 AM, Satish Balay <balay at mcs.anl.gov> wrote:

> On Fri, 16 Mar 2018, Matthew Knepley wrote:
>
> > On Thu, Mar 15, 2018 at 9:30 AM, Satish Balay <balay at mcs.anl.gov> wrote:
> >
> > > On Thu, 15 Mar 2018, Stefano Zampini wrote:
> > >
> > > > it seems we can no longer print the available configure options in
> master
> > > >
> > > > [szampini at localhost petsc]$ ./configure --help
> > > > ============================================================
> > > ===================
> > > >              Configuring PETSc to compile on your system
> > > >
> > > > ============================================================
> > > ===================
> > >
> > > Hm - its just taking a long time.. The trigger is this change.
> > >
> > > >>>>>>>>
> > > 603e7a67219475153ffb5c1693e4f8ee41f93af7 is the first bad commit
> > > commit 603e7a67219475153ffb5c1693e4f8ee41f93af7
> > > Author: Satish Balay <balay at mcs.anl.gov>
> > > Date:   Thu Mar 8 20:02:30 2018 -0600
> > >
> > >     rdict: attempt to workarround (not always reproduceable) error
> > >
> > >     *** Using PETSC_DIR=/home/sandbox/petsc/petsc.next
> > > PETSC_ARCH=arch-linux-opt-misc ***
> > >         No cached configure in RDict at /home/sandbox/petsc/petsc.
> > > next/arch-linux-opt-misc/lib/petsc/conf/RDict.db
> > >
> > > :040000 040000 654ea5265523cbcbb50b1104be63b5c7899b4478
> > > 01384d9a268f7c89859c64f9f091bee066d47cda M      config
> > > <<<<<<<<<<<<<<
> > >
> > > But I see script.py has code:
> > >
> > >    if self.hasHelpFlag():
> > >       self.argDB.readonly = True
> > >
> > >
> > > This code is not getting triggered '-h'/--help - I don't know if this
> is
> > > the correct behavior. Matt?
> > >
> >
> > Satish, I don't think this can be our fix. This writes to disk every
> time a
> > value is changed. We need to know
> > why the forced write at the end of configure was not working. This makes
> > everything forced.
>
> Sure - there has been random corruption of Rdict.db for many years and
> we haven't been able to track it down.  This change - with the force
> write at all times - is my attempt to avoid that corruption.
>
> I think in the prior discussion - you mentioned 'Rdict.db is not being
> used as per the original design intent'.
>
> If so - perhaps it should be removed? or written to only once - at the
> end? [or decouple 'make install' from it? I couldn't figure out this
> one.]


I agree. We should remove all code (about 2/3 of it) which does a hierarchy
of
communicating dicts (the original design). That would make everything
simple.
No threads, no parents, etc. We leave in the help the way we want it, types
for
args, etc. One thing its notably missing, and that PETSc Options are
missing,
is listing the thing that set the option (default, command line, code, env).

  Matt


>
> Satish
>
> >
> > I will look at why readonly is not working.
> >
> >   Matt
> >
> >
> > > Same slowdown with --with-packages-dir= [which behaves as help in one
> > > case - and no-help in the other case].. And we have this chane to fix
> > > its its misbehavior with --prefix
> > >
> > > 6ffe0648adde44dfc4a394659acd0d3e40176240
> > >
> > > Satish
> > >
> >
> >
> >
> >
>
>


-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener

https://www.cse.buffalo.edu/~knepley/ <http://www.caam.rice.edu/~mk51/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20180316/b04ae3b4/attachment-0001.html>