Proposed Browser Bookmark Sharing Standard.

[Marco Pesenti Gritti]

On Sat, 2003-06-28 at 05:59, Biju Chacko wrote:
> On 27 Jun 2003 13:32:13 +0200, Christophe Fergeau wrote:
> 
> > Writing standards for the sake of it doesn't seem like a really useful
> > thing, especially when the various browsers are already reasonably
> > interoperable. 
> 
> I agree with you wholeheartedly. Standards *must* serve a purpose.
> 
> IMO, browsers are insufficiently interoperable. I believe that users
> should be able to switch browsers transparently, without having to go
> through an intermediate step of 'importing bookmarks'. That is
> housekeeping that the system ought to be doing for him.
> 
> We need to fix the current view that bookmarks are the browser's data.
> They aren't -- they're the user's data.
> 
> I have nothing invested in this particular method of sharing data. If
> the consensus is that it sucks then it sucks. I wouldn't want that to
> obscure the underlying objective of transparent interoperability.
> 
> My main aim of posting it was to get folks who understand the problem
> better than I to start talking about it.

I've been doubious about posting my opinion because epiphany bookmarks
are quite different from traditional implementations. Anyway, I hope
it's useful.

> My questions are:
> 
> 1. Is what I am suggesting desirable and practical?

In general I think it's better to store bookmarks (and history) data in
a single file. Unlike applications you can have a  lot of bookmarks
installed, a file for each bookmarks would likely give performance
problems. Also, a lot of files make sharing bookmarks (with other users
or with other computers) problematic.

Going more in epiphany specific, I dont think we would be able to use
the specification because it set an user interface implementation.
I dont think the spec should assume that a bookmarks set is represented
in the interface by a hierarchical menu. That's true for most bookmarks
systems but for example it's not for Safari bookmarks where you can
select which bookmarks goes in menu.

Epiphany bookmarks are not stored in a hierarchy but are associated to
metadata. This is the rdf description of one bookmark:

 <item rdf:about="http://www.0xdeadbeef.com/html/2003/06/">
    <title>Blizzard blog</title>
    <link>http://www.0xdeadbeef.com/html/2003/06/</link>
    <dc:subject>News</dc:subject>
    <dc:subject>People</dc:subject>
  </item>

So
1 You can have multiple subjects unlike in a folder implementation where
you can have just one folder
2 You cant have the equivalent of nested subfolders.

I think a freedesktop bookmarks spec should just be a description of a
bunch of url and their properties because:

- it would allow different bookmarks system (with different interface
needs) to share it
- it would not set an ui interface implementation that we could not like
anymore in the future
- it would allow also other type of apps (not browsers) to use that data
- it keep the hierachy concept (which IHMO is a shame for usability) out
of the data format while still allowing a hierachical user interface

IHMO using .desktop files as description could work well with more
database/metadata like file systems, but in current context they would
give more problems then benefits.

> 2. Is it really worth the effort?

Personally I think bookmarks import is enough for the changing browser
user case. I think a common spec would be a lot more interesting for
tasks like:
- share bookmarks with other people
- use the same set of bookmarks with home and work box
- ability to access and modify bookmarks from other desktop apps

Marco