[petsc-dev] broken petsc-dev related mesh?

[Matthew Knepley]

On Sat, Mar 12, 2011 at 12:07 PM, Jed Brown <jed at 59a2.org> wrote:

> Matt moved dm/mesh to dm/impls/mesh, but did not update the makefile. Fix
> pushed.
>
> I would like to encourage running config/cmakegen.py --verbose occasionally
> since it shows stale things in the source tree, and stale makefiles. Some
> cases:
>
> src/sys/draw/impls/x/xvgrab.c has been both removed from the makefile and
> has #if 0 around its contents. Was the latter just to appease builder.py
> which (at that time, not sure now) was not checking makefiles? cmakegen.py
> parses the makefile to determine what to build. One reason to prefer that
> model is that it becomes easy to list those files which are not being built
> at all.
>

I still do not like this. I would prefer building everything in the
directory. I do not like sticking information in the makefile. I can
make builder do whatever we decide, since it only matter on the direct pass.


> src/sys/adic/adic/alloc.c is not a C source file at all, it hasn't been
> touched since it was first checked in in 2005.
>
> src/dm/impls/mesh/examples is not declared in the makefile, it would be
> consistent to move these to src/dm/examples, but many of the examples depend
> on SNES which is a higher level concept than DM, in which case they don't
> belong in DM at all.
>

Yes, they are all moving. It is going to take a while.


> The function DALocalToGlobal() vanished in the DMDA reorganization, the
> implementation is sitting in src/dm/impls/da/daltog.c but this file is not
> being compiled. Doesn't this functionality actually belong in DM, in which
> case it needs to be hooked back up?
>
>
> One final question/comment regarding Mercurial: when directories are moved
> or deleted (such as with cuda -> cusp naming), but the original contains
> generated files (*.html, *.d), the directory is not removed from the file
> system. I could use "hg clean" to remove all of this, but that is a blunt
> instrument because it removes things I might want to keep (like example
> output and man pages for things I care about). Is there a way to
> automatically remove the now-meaningless directories? Also, what do people
> think about generating *.html somewhere outside the source tree? (It looks
> like clutter to me and then we have to use ack or give flags to grep when
> searching the tree.)
>

I have no problem having a relocatable tree got html.

   Matt

-- 
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
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20110312/998a2df0/attachment.html>