Espen Antonsen

That looks like a good workaround @chewi ,have you been using this in production?

great @mphalliday, thanks for the feedback. My tests so far looks good. Thanks for sharing @chewi

@daniel-refahi because software development is a difficult task of managing the importance of a myriads of feature improvements, bug fixes and other issues. If you want this feature then give...

What server are you using? I would assume that the tmp directory would be writable otherwise (Heroku is sortof read-only).

And it works when setting the tmp dir? I'm a bit confused with the error message if the problem is not being able to store to disk.

A full stack trace would help. Try changing this and see if that helps. https://github.com/rick/balder/commit/81af9e7eedef9cf5a1d676213f888d6cfee40fa7

Thanks, I'll look into a better solution.

The script was supposed to ignore existing photos. It uses 'Photo.find_or_initialize_by_path(path)'. I guess this should be changed as path is not really used for photos. Maybe using 'Photo.find_or_initialize_by_file(path)' would work....

photo.Keywords.split(", ") should work if I recall the format correctly ("tag1, tag2")

The photoscan feature was developed to work locally, not with S3. I have updated this as a feature request. I do not have time or the need for this feature...