sandbox targets

[seth vidal]

On Thu, 2005-11-03 at 21:33 -0600, Josh Boyer wrote:
> On Thu, 2005-11-03 at 20:07 -0500, Jeff Spaleta wrote:
> > On 11/3/05, seth vidal <skvidal at phy.duke.edu> wrote:
> > > So what do people think? Should we make it so?
> > 
> > I'd certaintly use that option for anything i submitted into the
> > system from this point just as a sanity check on functionality before
> > going to publish.
> 
> Ditto.  One could even use it for *gasp* putting test packages out there
> for bug fixes.  I really like this sandbox idea.
> 
> > 
> > One thing to watch out for... if sandbox builds are a very popular
> > feature...you might consider sandbox builds be given lower priority
> > instead of fifowith  to-be-published packages. I'd have to see a lot
> > of sandbox builds que up and clog the buildhosts. But this is
> > obviously a secondary issue which might never need to be addressed.
> 
> I think this is a valid point.  It's better to at least think about it
> now rather than try to fix it when the buildsys is over-whelmed.  This
> is a case where being proactive will probably work out better than
> reactive.
> 

I disagree - overthinking it and adding a bunch of scheduler priority
mechanisms in the queuer will just make more work. Let's go for low
hanging fruit and see if we actually have a problem at all.

-sv