[petsc-dev] broken builds

[Matthew Knepley]

On Tue, Aug 30, 2011 at 9:03 PM, Jed Brown <jedbrown at mcs.anl.gov> wrote:

> On Tue, Aug 30, 2011 at 15:56, Matthew Knepley <knepley at gmail.com> wrote:
>
>> Distros have LOTS of manpower
>>
>
> I'm not sure I agree. The major distros have lots of people involved in
> core aspects (a handful might even get paid), but they have tens of
> thousands of community-maintained packages. They have to make the job of
> packaging easy in order to have the community maintain so many packages. A
> lot of those packages do upstream updates more frequently than most
> scientific libraries.
>
> The part of the problem that is harder for PETSc is that there is more
> variability in the environment we are installing into. The distros can
> assume a certain set of base tools that work in a consistent way, where as
> we need to be very careful about depending on anything.
>
> Figuring out how to do a DESTDIR install is specific to a package, not
> specific to the environment in which the install is taking place, therefore
> it is no harder for us than for the (volunteer contributor to the) distros.
>

This is exactly my point. The manpower is in having individual package
maintainers, which is what we would need and not have.


> Sandboxing the compile/install or setting system permissions (e.g. via
> fakeroot) would be harder for PETSc to do (because it's different on
> different systems).
>

Yes, nightmare.

We need to have enough to make the system work, so we can get real work
done.

   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/20110830/6b4a4e5c/attachment.html>