Re: Building against testing [was Re: "Recompile with libc6 from testing" <-- doesn't help i think!]

["Christopher W. Curtis" <ccurtis@aet-usa.com>]

Daniel Jacobowitz wrote:
> On Sun, Sep 15, 2002 at 07:03:42PM -0400, Christopher W. Curtis wrote:
> > At your request, I have given this more thought.  At first what you said 
> > made sense, but upon scrutinization, it's a red herring.  My initial 
> > points, since edited out, stated:
> >
> > *) Developers should run unstable
> > *) Builds should be done against testing
> > *) Uploads are always into unstable
[...]
> > Further, we have to take into account my first and last points: 
> > Firstly, if the developer is running unstable, it should be a reasonable 
> > assumption that they have done /some/ semblance of testing (against 
> > unstable, by definition) before uploading it.  Secondly, you make it 
>
> You're missing something.  Library packages are in unstable.  They need
> to go into testing.  "Testing", right?  They need to be _tested_ first. 
> If new applications don't link against the new versions, in many cases,
> they will silently make it to testing without ever being tested.

Hi,

I'll not drag this out, but I don't think that unbuildable is a valid 
concern as long as the first point is adhered to.  Presumably, if the 
developer is running unstable and personally building against unstable, 
there aren't going to be linker errors.  However, I also realize that 
this is unenforceable, lest buildd build twice - once against unstable 
"just to be sure" and then against testing for expidition.  However, I 
will take exception to your statement that unstable is necessarily a 
venue for testing.  I would think that that is what 'testing' is for. 
Testing is the precursor for stable, and is where testing is done.  This 
is probably in part why testing is so popular (no, sorry, I have no 
numbers).  Unstable is unstable, but testing may be as well.  The hope 
is that unstable allows for catastrophoc breakage without endangering 
the entire testing regime, but not to act as a replacement for it. 
That's why it's called "testING", and not "testED".  I think that Debian 
generally refers to the latter as "stable".

Of course, the opinion expressed herein is my own, and not necessarily 
representative of anyone else in the world.

Christopher