[Yum-devel] [PATCH] oddball case where rpm coughed up an EMPTY provides entry.

James Antill james at fedoraproject.org
Thu May 13 15:59:02 UTC 2010


On Thu, 2010-05-13 at 11:17 -0400, Seth Vidal wrote:
> I think it is safe to ignore and skip empty provides. This code
> lets that happen and makes sure that string_to_prco_tuple raises
> properly
> ---
>  yum/depsolve.py |    4 ++++
>  yum/misc.py     |    2 ++
>  2 files changed, 6 insertions(+), 0 deletions(-)
> 
> diff --git a/yum/depsolve.py b/yum/depsolve.py
> index 17b2ce9..b8fa6ad 100644
> --- a/yum/depsolve.py
> +++ b/yum/depsolve.py
> @@ -855,6 +855,8 @@ class Depsolve(object):
>          for req in sorted(txmbr_reqs, key=self._sort_req_key):
>              if req[0].startswith('rpmlib('):
>                  continue
> +            if not req[0]:
> +                continue
>              if req in oldreqs and self.rpmdb.getProvides(*req):
>                  continue
>              
> @@ -897,6 +899,8 @@ class Depsolve(object):
>                  continue
>              if prov in newpoprovs:
>                  continue
> +            if not prov[0]: # if the provname is '' then just skip it (opal-3.4.2-1.fc10 :()
> +                continue

 NAK. I don't think we want to do this here. Much better to filter them
at:

sqlitesack.py:returnPrco
packages.py:_populatePrco

...then nothing will ever get confused.



More information about the Yum-devel mailing list