On Wed, Sep 09, 2015 at 02:43:59PM +0200, Michal Židek wrote:
> Hi,
>
> patch for ticket
>
https://fedorahosted.org/sssd/ticket/2673
> is in the attachment.
>
> Thanks.
> Michal
> From 7c454bc2a737be05068418a5eef7fe9446bb5fa8 Mon Sep 17 00:00:00 2001
> From: =?UTF-8?q?Michal=20=C5=BDidek?= <mzidek(a)redhat.com>
> Date: Wed, 9 Sep 2015 14:37:48 +0200
> Subject: [PATCH] util: Include disabled domains in link_forest_roots
>
> Ticket:
>
https://fedorahosted.org/sssd/ticket/2673
> ---
> src/db/sysdb_subdomains.c | 6 +++---
> src/tests/cmocka/test_utils.c | 3 +++
> src/util/domain_info_utils.c | 21 ++++++++++++++++++---
> src/util/util.h | 3 +++
> 4 files changed, 27 insertions(+), 6 deletions(-)
The patch looks good but whenever I see us adding more and more boolean
switches, I wonder if we should just use flags instead?
This:
get_next_domain_ex(d, GND_USE_DISABLED | GND_DESCEND);
Reads quite a bit easier to me than:
get_next_domain_ex(d, true, false);
Also, bonus point are acquired next time we add a new flag, because not
all callers of the function must be converted..
What do you think?
I think this is very good point. It will also give us implicit
boolean parameter value (if not set, it is automatically false)
which improves readability a lot. I wander if it is good to add
the _ex function in this case. Would you agree if I changed the
original get_next_domain to use flags? I know, it needs to be changed
on more places, but I think such small refactoring makes more
sence than adding _ex function with flags.
Or is this what you were saying? I am not sure because
you used the _ex function in your example of "nice"
usage.
Michal