just a very little refactoring when touching the function
Resolve primary group in addition to secondary groups, so we don't
longer require memberUid present in the primary group.
However, if the memberUid is present, it may happen that the group is
included in the list twice, which means that it will be included in the
sysdb filter twice. I think this to be a better solution than to go over
all secondary groups and test whether it is also user's primary group.
We've been hitting situations where the sysdb conversion failed.
Unfortunately, the current code doesn't include enough debugging info to
pinpoint the failing entries. This patch adds more DEBUG statements for
each processed entry.
Please find attached the list of the 10 patches for ding-libs.
These patches implement merge capabilities of the ding-libs and refine
parser to consistently deal with different parse modes.
Patches 1-5 are self contained patches that can be applied independently
from each other but in the given order.
After applying each of those the tree should build-able.
Patches 6-10 are related to each other and need to be viewed and applied
in relation. They are a logically split for an easier review one piece
of functionality .
Patch names and commit comments give an idea of what happens in each patch.
Most complicated patches are 6, 7 and 10.
Let me know if I missed something.
Sr. Engineering Manager for IdM portfolio
Red Hat Inc.
Looking to carve out IT costs?
I ran the new version of scan-build and checked the SSSD. There's a big
number of false positives, anything that follows this construct:
struct foostruct *f = NULL;
Triggers a warning that f->foomember is dereferencing NULL.
That said, clang also uncovered some real bugs. Attached are fixes for
bugs that are obvious, I will follow up with some others with people who
have more knowledge about the code in question.