[ovs-dev] [PATCH monitor_cond V6 01/11] ovsdb: create column index mapping between ovsdb row to monitor row

Liran Schour LIRANS at il.ibm.com
Tue Jun 7 05:21:07 UTC 2016


Ben Pfaff <blp at ovn.org> wrote on 06/06/2016 06:13:44 PM:

> On Mon, Jun 06, 2016 at 01:11:44PM +0300, Liran Schour wrote:
> > Ben Pfaff <blp at ovn.org> wrote on 01/06/2016 09:18:46 PM:
> > > This new columns_index_map means that it's no longer necessary to 
work
> > > so hard to find duplicates--rather, we can find a duplicate at the 
time
> > > we try to add a column: a new column is a duplicate if
> > > columns_index_map[column->index] != -1 at the time we try to add it. 
 It
> > > would probably be nicer to simply report the error at the time we 
add
> > > it, instead of adding a sort step and a re-indexing step.
> > 
> > Will change the code to find column duplication on column add.
> 
> When I kept reading in later patches, I realized that there is some
> value in sorting the columns because it makes it easier to compare two
> sets of columns for equality.  I don't know whether this affects when it
> makes sense to detect duplicates.
> 

Right. Changed the code that duplication of column is detected on 
ovsdb_add_column(). Sorting columns set and then comparing with other sets 
is down under ovsdb_monitor_add().




More information about the dev mailing list