On Mon, May 27, 2013 at 08:06:35PM +0200, Jakub Hrozek wrote:
On Fri, May 24, 2013 at 08:32:20AM +0200, Sumit Bose wrote:
> On Thu, May 23, 2013 at 04:39:23PM -0400, Simo Sorce wrote:
> > On Thu, 2013-05-23 at 22:20 +0200, Sumit Bose wrote:
> > >
> > > maybe it should be added that it is expected that there is a delimiter
> > > of one character not more or less.
> > >
> > The original code was able to cope with delimiters bigger than 1 char,
> > has that changed ?
>
> ah, you are right, I've just see 'int delim = 1;' in fill_pwent and
> friends but not that it is reset if more space is needed.
>
> bye,
> Sumit
I confess I wasn't quite sure about the code either when I opened it
after two weeks, so at least there is a unit test added in the latest
iteration of patches and also there is a new patch in the series that
attempts to make the flow easier by removing some "else" branches.
Thank you for the review, all your comments should be fixed now. New
patches are attached, but only patch #3 has changed and patch #4 is new.
There is a new utility function sss_fqdom_len() that is used instead of
simply strdup(domain) when formatting the name into a fully-qualified-name. I
haven't found any other places to fix, at least "git grep strlen | grep
dom"
doesn't show me any :)
Patches look good and worked in my test, even with weird fully qualified
names where long and short names are used together multiple times, so
ACK.
While testing I found an issue not related to the patches but important
for them to work properly. I opened
https://fedorahosted.org/sssd/ticket/1951 about it and will look how ti
fix it.
bye,
Sumit