[Freeipa-devel] [PATCH] 11 Fix unitialized attributes

[Martin Kosek]

On Thu, 2011-04-21 at 10:18 +0200, Jan Cholasta wrote:
> On 14.4.2011 12:58, Jan Cholasta wrote:
> > On 13.4.2011 15:21, Jan Cholasta wrote:
> >> On 13.4.2011 13:04, Martin Kosek wrote:
> >>> On Mon, 2011-04-11 at 15:07 +0200, Jan Cholasta wrote:
> >>>> Fixed unitialized attributes in ipalib.aci.ACI, ipalib.parameters.Data,
> >>>> ipalib.text.LazyText, ipaserver.install.service.Service and
> >>>> ipaserver.rpcserver.WSGIExecutioner.
> >>>>
> >>>
> >>> NACK. This patch breaks the build.
> >>>
> >>> $ make rpms
> >>> [snip]
> >>> if [ "" != "yes" ]; then \
> >>> ./makeapi --validate; \
> >>> fi
> >>> Traceback (most recent call last):
> >>> File "./makeapi", line 281, in<module>
> >>> sys.exit(main())
> >>> File "./makeapi", line 259, in main
> >>> api.finalize()
> >>> File "/home/mkosek/freeipa/ipalib/plugable.py", line 574, in finalize
> >>> self.__do_if_not_done('load_plugins')
> >>> File "/home/mkosek/freeipa/ipalib/plugable.py", line 384, in
> >>> __do_if_not_done
> >>> getattr(self, name)()
> >>> File "/home/mkosek/freeipa/ipalib/plugable.py", line 526, in
> >>> load_plugins
> >>> self.import_plugins('ipalib')
> >>> File "/home/mkosek/freeipa/ipalib/plugable.py", line 564, in
> >>> import_plugins
> >>> raise e
> >>> AttributeError: locked: cannot set Str.re to None
> >>> make: *** [version-update] Error 1
> >>
> >> Fixed.
> >
> > NACK, made a stupid mistake. Should be ok now.
> 
> More elegant solution, per Martin's request.

ACK. Good, this is much more elegant solution.
Pushed to master, ipa-2-0.

Honza, one more thing. Please, include patch version into it's filename,
e.g. freeipa-jcholast-11-4-uninitialized-attributes.patch. This will
prevent confusion and will also make it easier to get differences
between patch versions. I won't have to rename it manually myself.

This is the last patch for pylint errors, our code is pylint-clean (with
python-rhsm installed). I think we can integrate "make lint" to our
build process as we talked earlier.

I have one more suggestion regarding pylint implementation (as we
discussed). When a python package is missing (python-rhsm at the moment)
pylint produces false positives without any further explanation - until
the missing package is installed. It would be great to have this
covered.

Martin