On (06/09/16 21:38), Fabiano Fidêncio wrote:
On Tue, Sep 6, 2016 at 8:49 PM, lslebodn
> lslebodn commented on a pull request
> IMHO, it might be better to close this PR.
> If we decide to dor support for libini_config < 1.1 or 1.2
> then it will be a different patch anyway. @see my previous comment
Please, take a look on Jakub's comment and see what he proposed.
Do you mean:
jhrozek commented on a pull request
> - Ubuntu 12.04 LTS and older
12.04 is often used (for example travis-ci still offers only this
distro) and ends its life in 2017.
On one hand, it's unlikely that users of LTS distributions will run
master, they will probably only run sssd-1-13 or some PPAs, on the other
hand, I don't see too much value except for cleaner code.
So all in all I suggest we nack this patchset for now and push it when
Ubuntu 12.04 goes EOL.
Does that sound like a reasonable compromise?
If we decide to drop support for older versions of libini_config
then we should also remove unnecessary wrappers in src/util/sss_ini.c.
and replace sss_ini* functions with direct usage of ini_*.
Therefore this patchset will need to be changed.
IMHO, it will be much simpler to create new patch-set then.
Anyways, I'm fine with closing this PR and opening a new one when
can drop support to libini_config < 1.3.
Just for the record, from this whole discussion I could see that
dropping support to augeas in order to use libini is something that
shouldn't and is not going to happen any time soon
I do not understand how is
this patch-set related to replacing augeas
a) augeas is an optional dependency
pkglib_LTLIBRARIES += libsss_config.la
b) replacing augeas with libini_config requires minimal version 1.2
which is already optionally used in master.
c) IIRC discussion from devel meeting. Some developers prefer
to drop feature for manipulation of sssd.conf and write from
scratch if there will be new requirements. Current version is very
as well and for the
very same reasons that this patch wasn't accepted. We could, in the
best (or worst?) case scenario support/depend on both due to
compatibility with elder systems, but I can't see any advantage on
I do not suggest to support all features in old systems.
But I would like to support at least minimal feature set (ldap+krb5).
And parsing configuration file is required for minimal feature set :-)