bruns added a comment.
On the topic of "multiple values from one extractor" vs "multiple values from different extractors": I think, when this problem arrives, it should be solved by using one PropertyMap per extractor and explicitly merging these. 1. For per-extractor unique properties, this is just the union 2. For repeated property values, just use the value once 3. For contradicting values, this has to be dealed with on a per-property basis. This avoids the extra storage in case of duplicated values, keeps the (recurrent!) cost and complexity of merging away from the consumers, and allows better handling or merging, as "dumb" merging of properties already loses information (e.g. where the information originated from). REPOSITORY R293 Baloo REVISION DETAIL https://phabricator.kde.org/D19087 To: bruns, #baloo, #frameworks, ngraham, poboiko, astippich Cc: kde-frameworks-devel, gennad, domson, ashaposhnikov, michaelh, astippich, spoorun, ngraham, bruns, abrahams
