31. ledna 2016 15:04:48 CET, Anton Aylward <opensuse@antonaylward.com> napsal:
On 01/31/2016 04:53 AM, Vojtěch Zeisek wrote:
I do not trust any tool saving comments outside commented files and specifically I do not trust KDE semantic tools in this way because of various problems and technological instability. My personal impression, but no, thanks. Good for searching, not for storing extra information.
Indeed! And ultimately its not about trust. it's an operational thing.
Storing any such information outside the file means it can be 'left behind'. it means it is platform dependent, the platform being not simply Linux vs OSX vs Windows, but which DM to use, which applicaiton to use.
That's the whole point of embedded metatdata.
Some file systems even have 'forks' in the file, the file, rather than being an array of bytes, has branches, is itself a database. if effect this is what the exif data is. The 'header' has a number of fields, some of which are pointers. They may point to the actual data. In the case of many video files there may also be a pointer to an embedded snapshot.
Now here is the kicker: There are purists out there who say you should !NEVER! edit the exif of the sources images. they say this about the RAW images from cameras and even from scanners and many tiffs, jpgs and gifs that are produced by proprietary systems. They claim that only the vendor really knows what the fields and pointers are, regardless of the standards in this area, regardless of our skills at reverse engineering.
Ultimately, those header tables are not much different from the header tables of many program load images and their link tables. We've over half a century of dealing with linkage editing. We've almost as much experience in reverse engineering even when the vendors/manufacturers don't publish data, and many of the video/camera manufacturers DO publish their specs.
And we have plenty of examples of the reverse engineering 'hackers' ending up knowing more than the manufacturers and finding bugs and inconsistencies!
Its one thing to apply j.random.edit to the exif, as it is to any file, but there are some exif/IPC field s that are clearly there input/update: comment, author, description, copyright. In the limiting case, saying that a copyright stamp should be in an external database that has no inherent association with the image is, in my opinion, lunacy! Ask someone who has had their image or recording pirated and relied on the EMBEDDED copyright information to make their case.
I've had this argument with the authors of 'darktable'. They feel its adequate to leave the original files alone and have the extra stuff in a XML sidebar file, and then their software add it when the generate a jpg or gif. They don't see that they, in turn, are applying an edit and generating their own table and set of pointers in this image. And of course its "documented" because their code is open source.
The idea that authorship and copyright should be applied as early as possible and in a way that is tightly associated with the ORIGINAL image (rather than derivative works) seems to escape them, but then this gets down to being a legal issue rather than a technological one. Having the copyright claim in a sidebar file or in an external database raises many legal issues.
Its not as if we are talking about editing any of the technical fields, the EV, aperture, shutter. In fact editing those fields would not involve restructuring and altering length of fields and hence pointers.
Yes, yes, yes! And how is it with that interoperability then? Metadata inside files are the most portable and easily managed. -- Vojtěch Zeisek http://trapa.cz/cs -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org