On Fri, May 17, 2013 at 10:29 AM, Bernardo Dal Seno <[email protected]>wrote:

> On 17 May 2013 08:18, Thomas Thrainer <[email protected]> wrote:
> >
> >
> >
> > On Thu, May 16, 2013 at 5:17 PM, Bernardo Dal Seno <[email protected]>
> > wrote:
> >>
> >> On 16 May 2013 12:14, Bernardo Dal Seno <[email protected]> wrote:
> >> > On 16 May 2013 10:23, Thomas Thrainer <[email protected]> wrote:
> >> >> LUNetwork* and associated helper functions are extracted to
> network.py.
> >> >>
> >> >> Signed-off-by: Thomas Thrainer <[email protected]>
> >> >> ---
> >> >>  Makefile.am            |   3 +-
> >> >>  lib/cmdlib/__init__.py | 714
> >> >> +-----------------------------------------------
> >> >>  lib/cmdlib/common.py   |  25 ++
> >> >>  lib/cmdlib/network.py  | 718
> >> >> +++++++++++++++++++++++++++++++++++++++++++++++++
> >> >>  4 files changed, 749 insertions(+), 711 deletions(-)
> >> >>  create mode 100644 lib/cmdlib/network.py
> >> >
> >> > LGTM, thanks.
> >>
> >> Not really, sorry. This breaks the ordering:
> >>
> > Just a quick question: Are all lists (imports, in makefiles, etc.) always
> > sorted alphabetically? I actually sorted the files in the makefile on
> > purpose, going from "important" to "less important". The end result is
> (see
> > comments for my reasoning):
> >
> > lib/cmdlib/__init__.py \  # main entry point, thus at the top
> > lib/cmdlib/common.py \  # common and base stuff, needed by everything ->
> top
> > lib/cmdlib/base.py \
> > lib/cmdlib/cluster.py \  # ordering here: cluster (big) -> group (part of
> > cluster) -> node (part of group) -> instance (on node)
> > lib/cmdlib/group.py \
> > lib/cmdlib/node.py \
> > lib/cmdlib/instance.py \  # the parts of instance are a bit random, true
> > lib/cmdlib/instance_storage.py \
> > lib/cmdlib/instance_migration.py \
> > lib/cmdlib/instance_operation.py \
> > lib/cmdlib/instance_query.py \
> > lib/cmdlib/instance_utils.py \   # "private" helper functions for
> instance.
> > In my reasoning, private (detailed) stuff goes a the end ;-)
> > lib/cmdlib/backup.py \  # here comes the somewhat unsorted rest, which
> does
> > not quite align to a "physical" entity (cluster, group, node, instance)
> > lib/cmdlib/query.py \
> > lib/cmdlib/operating_system.py \
> > lib/cmdlib/tags.py \
> > lib/cmdlib/network.py \
> > lib/cmdlib/misc.py \
> > lib/cmdlib/test.py  # tests are the least important, thus at the end
> >
> > So, the question is, is it a guideline to always sort such lists
> > alphabetically, or are other reasoning permitted as well? Does my
> reasoning
> > make sense to anybody except me :)?
>
> Alphabetical ordering is clear and it means the same to everybody.
> Important is a fuzzy concept, it makes things difficult if you want to
> see if a file is present in the list, and it's not very maintainable:
> You want to change the Makefile when you move an important function
> from one file to another?
>
> The other lists in the Makefile (maybe not all of them) are sorted
> that way, so I won't deviate for this one. But feel free to ask
> others' opinion.
>
>
No problem, I'll re-sort the makefile. Is it ok if I just leave the order
as it is in the individual patches, and send a new patch for the reordering
at the end of this series? I'd like to avoid another round of rebasing :-).


> Bernardo
>



-- 
Thomas Thrainer | Software Engineer | [email protected] |

Google Germany GmbH
Dienerstr. 12
80331 München

Registergericht und -nummer: Hamburg, HRB 86891
Sitz der Gesellschaft: Hamburg
Geschäftsführer: Graham Law, Katherine Stephens

Reply via email to