On Thu, 15 Mar 2012 18:42:00 +0000, Mark Walters wrote: > This series changes the --no-exclude options to count, search and show > to --exclude=(true|false|flag). It also changes the default to true > rather than flag for speed and for uncluttered command line output. > > This series replaces the series > id:"1330779918-28024-1-git-send-email-markwalters1009@gmail.com". Jani > suggested the --exclude= rather than --with-excluded approach as being > both clearer and easier to extend. > > It is intend to apply on top of the bugfix series > id:"1331728014-32698-1-git-send-email-markwalters1009@gmail.com". And indeed it does. I've tested and reviewed and it looks and works great. > There should be no significant logic change (i.e. the logic in the lib > is unchanged) but the defaults for the command lines are different. I fully support this change. The new logic makes sense to me and is a good compromise between usability and flexibility. > The general idea is that when complete threads are requested then we > always return complete threads including any messages matching the > exclude tags (where appropriate these are flagged excluded), but we > may not return threads that only match in excluded messages. > > The rationale is that it is awkward to deal with a thread with > "missing" messages. Agreed. I should note that for all uses of excludes that I can think of one would not expect to be excluding messages from threads that one would otherwise be keeping. So I certainly have no problem not excluding Thanks again for working on this, Mark. Your efforts are much appreciated. This is nice new functionality. jamie.