[XESAM] Minutes of meeting 2007-05-15

[Joe Shaw]

Hi,

On 5/18/07, Mikkel Kamstrup Erlandsen <mikkel.kamstrup at gmail.com> wrote:
> Ok, so taking a conrete case, a "document" file type with hit type "contact"
> could be a vcard stored somewhere on disk?

Close, but not exactly.  For a vcard file stored in the user's file
system, the hit type would be "file" because it's a file on disk, and
not an abstract "contact" pulled from some contact DB like Evolution
Data Server.  File type would be "contact" because that is an accurate
description of what the file itself is.

In the context of file types in Beagle, "document" means something
like an office document. OOo docs, MS Office docs, and PDFs are all
examples of things that fall into our "document" file type.

> It took my a sec to grok the distinction between hit type and file type. If
> I understand correctly hit type could also be called "present as", and yes
> present-as and file-type are independent the way you describe it. However
> this is not the way I would abstract things. I think a category tree like
> http://www.grillbar.org/xesam/object-graph.png would be
> more natural. In this way both MailAttachment and files in a tarball are
> both EmbeddedObjects in a natural way.

I'm trying to merge this paragraph with the image URL you gave... does
this mean that for each possible EmbeddedObject type, the entire
Object hierarchy would be duplicated?  Ie:

Object -> (Video, Audio, Person, etc.)
EmbeddedObject -> MailAttachment -> (Video, Audio, Person, etc.)
EmbeddedObject -> ArchivedFile -> (Video, Audio, Person, etc.)

If so, I'm not crazy about this at all.  It feels a lot more natural
to consider the concepts of what category the file is (file type) and
where it comes from (hit type) as two independent pieces of
information, and essentially treat them as interfaces implemented by a
document.

Joe