[Shotwell] any news on multi-tag select?

[Dougie Nisbet]

On 06/04/2012 19:49, Colin Law wrote:
> On 6 April 2012 19:26, Adam Dingle<adam at yorba.org>  wrote:
>> On 04/06/2012 07:47 AM, Adam Dingle wrote:
>>
>> On 04/06/2012 07:42 AM, Dougie Nisbet wrote:
>>
>> On 06/04/2012 15:30, Colin Law wrote:
>>
>> So if Dougie goes back to a previous database and uses that with released
>> 0.12 his tags should not get flattened this time? Colin
>>
>>
>> No, I would certainly expect not.
>>
>> That what I was wondering but I don't think it works like that. If I read my
>> 0.11.6 db with 0.12.1 they're flattened.
>>
>>
>> Really?  That would be a serious bug.  Could you send a copy of your 0.11.6
>> db (~/.shotwell/data/photo.db) to shotwell at yorba.org so we can investigate?
>> 0.12.1 should not be flattening hierarchical tags.
>>
>>
>> Dougie,
>>
>> thanks for sending us your Shotwell database and a video demonstrating the
>> problem.  Here's what I think is going on.  When you were burned by bug
>> http://redmine.yorba.org/issues/4925 in the Shotwell prerelease, Shotwell
>> flattened your hierarchical tags - and then wrote those flattened tags to
>> the photo files (you must have had metadata writing enabled).  Even if you
>> restore an old copy of your photo.db, the flattened tags are still in the
>> photo files.  So when Shotwell starts up, it reads those tags from the
>> files, adds them to its database and displays them in the sidebar.  This is
>> entirely expected behavior.
> That does not seem to fit the facts as a bit earlier Dougie said that
> he had moved .shotwell out of the way so that it would reimport the
> files (in place if I understood correctly) and that this is
> re-building the tags.  That suggests that the photos have the
> hierarchical tags still in place.

I think the bottom line is that the 'flattened tags' information is 
stored in the image files themselves, and whether I like it or not, or 
tell it or not, shotwell parses the image library on startup and 
attempts to make sense of what it finds. The "scan library on startup" 
(regardless) has come up as an issue on the mailing list a few times.  I 
suspect my images contain one unholy mess of exif tag data that is a 
tramsmash of tags that have resulted from using the various pre-release 
shotwells.

>>   you have two choices.

There might be a third option. If my db is ok, and the problem is 
shotwell parsing the library, then if I nuke the exif tags on all my 
images and allow shotwell to re-write them from its db, that might work? 
e.g. But running  'exiftool -all= "${fname}"' on all images might work?

Dougie