> I'm not sure now what you mean by tagging, but I feel obliged to
> emphasize that in the Camlistore model, the permanode is the
> representation of your object. It's the only entity that gets
> (mutable) attributes (through claims), which is why it is the one
> being tagged. So if you think the two images are different objects
> then you assign each of them to a different permanode, that you will
> both tag however you like. If not, then there's nothing to worry
> about, is there?

Okay, so let me give you an example to illustrate what I'm talking
about. If I have two large backups of all of my images, and I import
both of them into camlistore I will end up with more than one copy of
a "file" that represents the same image. This isn't a _huge_ problem
(though because of the schema it becomes a problem), but it means that
I have to pick one of the images to add tags to. I cannot add tags to
both (otherwise you get two hits when you do a search) and I cannot
add a tag to the image itself (as in, a permanode that points to just
the bytes, not the file metadata). So the metadata has to be
duplicated or asymmetric (neither of which is good).

Now, you could argue that I should've deduped my files before
importing them, but isn't that against the point of camlistore?

My biggest beef with Camlistore at the moment is that everything is
treated like a file. Even tweets. They're all treated as files, even
though the point is that it's an _object_ store. Why aren't we storing
the raw data (meaning the whole image and the tags associated with it
-- ie permanode -> bytes -> image contents) as objects and then adding
the file abstraction *on top* of that. I get that you can do it
manually, but I could do all of this manually with S3 if I wanted to
-- the point of camlistore is that this is all supported upfront.

-- 
Aleksa Sarai (cyphar)
www.cyphar.com

-- 
You received this message because you are subscribed to the Google Groups 
"Camlistore" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to