On Fri, Jun 06, 2014 at 09:28:46AM +0200, Vratislav Podzimek wrote:
On Thu, 2014-06-05 at 07:35 -0700, Brian C. Lane wrote:
> On Thu, Jun 05, 2014 at 12:56:29PM +0200, Vratislav Podzimek wrote:
> > Users are responsible for package selection in case they explicitly excluded
one
> > of storage/bootloader-required packages.
> >
> > Signed-off-by: Vratislav Podzimek <vpodzime(a)redhat.com>
> > ---
> > pyanaconda/install.py | 6 ++++--
> > 1 file changed, 4 insertions(+), 2 deletions(-)
> >
> > diff --git a/pyanaconda/install.py b/pyanaconda/install.py
> > index 2bdbbd1..050c832 100644
> > --- a/pyanaconda/install.py
> > +++ b/pyanaconda/install.py
> > @@ -197,8 +197,10 @@ def doInstall(storage, payload, ksdata, instClass):
> > if not ksdata.bootloader.disabled:
> > packages += storage.bootloader.packages
> >
> > - # don't try to install packages from the install class' ignored
list
> > - packages = [p for p in packages if p not in instClass.ignoredPackages]
> > + # don't try to install packages from the install class' ignored
list and the
> > + # explicitly excluded ones (user takes the responsibility)
> > + packages = [p for p in packages
> > + if p not in instClass.ignoredPackages and p not in
ksdata.packages.excludedList]
> > payload.preInstall(packages=packages, groups=payload.languageGroups())
> > payload.install()
> >
> > --
> > 1.9.3
>
> When this inevitably leads to someone excluding something we actually
> use are the errors going to be clear enough that they won't file a bug?
Well, I cannot think of any way how to achieve that. We already have
the --nocore option for %packages that may lead to such errors (I've
encountered one myself yesterday) and there's no special handling of
such cases.
Thus I'd like to go with this patch and then try to resolve the general
issue with explicitly excluded packages leading to errors somehow
systematically. I was thinking about adding ksdata.packages.excludedList
Oh, certainly, sorry, I didn't mean for that comment to be an ACK. Just
something to think about.
to our exception handling, but it's probably impossible to
determine if
the missing utility is a part of some excluded package or not. Any
ideas?
The only thing I can come up with right now is to make sure that when a
OSError is hit that we log the full cmdline we were trying to run.
They're logging in program.log but it may prevent bugs being files if it
were part of the traceback.
--
Brian C. Lane | Anaconda Team | IRC: bcl #anaconda | Port Orchard, WA (PST8PDT)